log_event_server.cc 282 KB
Newer Older
1
/*
Marko Mäkelä's avatar
Marko Mäkelä committed
2
   Copyright (c) 2000, 2019, Oracle and/or its affiliates.
Marko Mäkelä's avatar
Marko Mäkelä committed
3
   Copyright (c) 2009, 2022, MariaDB
4 5 6 7 8 9 10 11 12 13 14 15

   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
Marko Mäkelä's avatar
Marko Mäkelä committed
16
   Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1335  USA */
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48


#include "mariadb.h"
#include "sql_priv.h"

#ifdef MYSQL_CLIENT
#error MYSQL_CLIENT must not be defined here
#endif

#ifndef MYSQL_SERVER
#error MYSQL_SERVER must be defined here
#endif

#include "unireg.h"
#include "log_event.h"
#include "sql_base.h"                           // close_thread_tables
#include "sql_cache.h"                       // QUERY_CACHE_FLAGS_SIZE
#include "sql_locale.h" // MY_LOCALE, my_locale_by_number, my_locale_en_US
#include "key.h"        // key_copy
#include "lock.h"       // mysql_unlock_tables
#include "sql_parse.h"  // mysql_test_parse_for_slave
#include "tztime.h"     // struct Time_zone
#include "sql_load.h"   // mysql_load
#include "sql_db.h"     // load_db_opt_by_name
#include "slave.h"
#include "rpl_rli.h"
#include "rpl_mi.h"
#include "rpl_filter.h"
#include "rpl_record.h"
#include "transaction.h"
#include <my_dir.h>
#include "sql_show.h"    // append_identifier
Marko Mäkelä's avatar
Marko Mäkelä committed
49
#include "debug_sync.h"  // debug_sync
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
#include <mysql/psi/mysql_statement.h>
#include <strfunc.h>
#include "compat56.h"
#include "wsrep_mysqld.h"
#include "sql_insert.h"

#include <my_bitmap.h>
#include "rpl_utility.h"
#include "rpl_constants.h"
#include "sql_digest.h"
#include "zlib.h"


#define log_cs  &my_charset_latin1


#if defined(HAVE_REPLICATION)
static int rows_event_stmt_cleanup(rpl_group_info *rgi, THD* thd);

static const char *HA_ERR(int i)
{
  /* 
    This function should only be called in case of an error
    was detected 
   */
  DBUG_ASSERT(i != 0);
  switch (i) {
  case HA_ERR_KEY_NOT_FOUND: return "HA_ERR_KEY_NOT_FOUND";
  case HA_ERR_FOUND_DUPP_KEY: return "HA_ERR_FOUND_DUPP_KEY";
  case HA_ERR_RECORD_CHANGED: return "HA_ERR_RECORD_CHANGED";
  case HA_ERR_WRONG_INDEX: return "HA_ERR_WRONG_INDEX";
  case HA_ERR_CRASHED: return "HA_ERR_CRASHED";
  case HA_ERR_WRONG_IN_RECORD: return "HA_ERR_WRONG_IN_RECORD";
  case HA_ERR_OUT_OF_MEM: return "HA_ERR_OUT_OF_MEM";
  case HA_ERR_NOT_A_TABLE: return "HA_ERR_NOT_A_TABLE";
  case HA_ERR_WRONG_COMMAND: return "HA_ERR_WRONG_COMMAND";
  case HA_ERR_OLD_FILE: return "HA_ERR_OLD_FILE";
  case HA_ERR_NO_ACTIVE_RECORD: return "HA_ERR_NO_ACTIVE_RECORD";
  case HA_ERR_RECORD_DELETED: return "HA_ERR_RECORD_DELETED";
  case HA_ERR_RECORD_FILE_FULL: return "HA_ERR_RECORD_FILE_FULL";
  case HA_ERR_INDEX_FILE_FULL: return "HA_ERR_INDEX_FILE_FULL";
  case HA_ERR_END_OF_FILE: return "HA_ERR_END_OF_FILE";
  case HA_ERR_UNSUPPORTED: return "HA_ERR_UNSUPPORTED";
  case HA_ERR_TO_BIG_ROW: return "HA_ERR_TO_BIG_ROW";
  case HA_WRONG_CREATE_OPTION: return "HA_WRONG_CREATE_OPTION";
  case HA_ERR_FOUND_DUPP_UNIQUE: return "HA_ERR_FOUND_DUPP_UNIQUE";
  case HA_ERR_UNKNOWN_CHARSET: return "HA_ERR_UNKNOWN_CHARSET";
  case HA_ERR_WRONG_MRG_TABLE_DEF: return "HA_ERR_WRONG_MRG_TABLE_DEF";
  case HA_ERR_CRASHED_ON_REPAIR: return "HA_ERR_CRASHED_ON_REPAIR";
  case HA_ERR_CRASHED_ON_USAGE: return "HA_ERR_CRASHED_ON_USAGE";
  case HA_ERR_LOCK_WAIT_TIMEOUT: return "HA_ERR_LOCK_WAIT_TIMEOUT";
  case HA_ERR_LOCK_TABLE_FULL: return "HA_ERR_LOCK_TABLE_FULL";
  case HA_ERR_READ_ONLY_TRANSACTION: return "HA_ERR_READ_ONLY_TRANSACTION";
  case HA_ERR_LOCK_DEADLOCK: return "HA_ERR_LOCK_DEADLOCK";
  case HA_ERR_CANNOT_ADD_FOREIGN: return "HA_ERR_CANNOT_ADD_FOREIGN";
  case HA_ERR_NO_REFERENCED_ROW: return "HA_ERR_NO_REFERENCED_ROW";
  case HA_ERR_ROW_IS_REFERENCED: return "HA_ERR_ROW_IS_REFERENCED";
  case HA_ERR_NO_SAVEPOINT: return "HA_ERR_NO_SAVEPOINT";
  case HA_ERR_NON_UNIQUE_BLOCK_SIZE: return "HA_ERR_NON_UNIQUE_BLOCK_SIZE";
  case HA_ERR_NO_SUCH_TABLE: return "HA_ERR_NO_SUCH_TABLE";
  case HA_ERR_TABLE_EXIST: return "HA_ERR_TABLE_EXIST";
  case HA_ERR_NO_CONNECTION: return "HA_ERR_NO_CONNECTION";
  case HA_ERR_NULL_IN_SPATIAL: return "HA_ERR_NULL_IN_SPATIAL";
  case HA_ERR_TABLE_DEF_CHANGED: return "HA_ERR_TABLE_DEF_CHANGED";
  case HA_ERR_NO_PARTITION_FOUND: return "HA_ERR_NO_PARTITION_FOUND";
  case HA_ERR_RBR_LOGGING_FAILED: return "HA_ERR_RBR_LOGGING_FAILED";
  case HA_ERR_DROP_INDEX_FK: return "HA_ERR_DROP_INDEX_FK";
  case HA_ERR_FOREIGN_DUPLICATE_KEY: return "HA_ERR_FOREIGN_DUPLICATE_KEY";
  case HA_ERR_TABLE_NEEDS_UPGRADE: return "HA_ERR_TABLE_NEEDS_UPGRADE";
  case HA_ERR_TABLE_READONLY: return "HA_ERR_TABLE_READONLY";
  case HA_ERR_AUTOINC_READ_FAILED: return "HA_ERR_AUTOINC_READ_FAILED";
  case HA_ERR_AUTOINC_ERANGE: return "HA_ERR_AUTOINC_ERANGE";
  case HA_ERR_GENERIC: return "HA_ERR_GENERIC";
  case HA_ERR_RECORD_IS_THE_SAME: return "HA_ERR_RECORD_IS_THE_SAME";
  case HA_ERR_LOGGING_IMPOSSIBLE: return "HA_ERR_LOGGING_IMPOSSIBLE";
  case HA_ERR_CORRUPT_EVENT: return "HA_ERR_CORRUPT_EVENT";
  case HA_ERR_ROWS_EVENT_APPLY : return "HA_ERR_ROWS_EVENT_APPLY";
127
  case HA_ERR_PARTITION_LIST : return "HA_ERR_PARTITION_LIST";
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154
  }
  return "No Error!";
}


/*
  Return true if an error caught during event execution is a temporary error
  that will cause automatic retry of the event group during parallel
  replication, false otherwise.

  In parallel replication, conflicting transactions can occasionally cause
  deadlocks; such errors are handled automatically by rolling back re-trying
  the transactions, so should not pollute the error log.
*/
static bool
is_parallel_retry_error(rpl_group_info *rgi, int err)
{
  if (!rgi->is_parallel_exec)
    return false;
  if (rgi->speculation == rpl_group_info::SPECULATE_OPTIMISTIC)
    return true;
  if (rgi->killed_for_retry &&
      (err == ER_QUERY_INTERRUPTED || err == ER_CONNECTION_KILLED))
    return true;
  return has_temporary_error(rgi->thd);
}

155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
/**
  Accumulate a Diagnostics_area's errors and warnings into an output buffer

    @param errbuf       The output buffer to write error messages
    @param errbuf_size  The size of the output buffer
    @param da           The Diagnostics_area to check for errors
*/
static void inline aggregate_da_errors(char *errbuf, size_t errbuf_size,
                                       Diagnostics_area *da)
{
  const char *errbuf_end= errbuf + errbuf_size;
  char *slider;
  Diagnostics_area::Sql_condition_iterator it= da->sql_conditions();
  const Sql_condition *err;
  size_t len;
  for (err= it++, slider= errbuf; err && slider < errbuf_end - 1;
       slider += len, err= it++)
  {
    len= my_snprintf(slider, errbuf_end - slider,
                     " %s, Error_code: %d;", err->get_message_text(),
                     err->get_sql_errno());
  }
}

179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198

/**
   Error reporting facility for Rows_log_event::do_apply_event

   @param level     error, warning or info
   @param ha_error  HA_ERR_ code
   @param rli       pointer to the active Relay_log_info instance
   @param thd       pointer to the slave thread's thd
   @param table     pointer to the event's table object
   @param type      the type of the event
   @param log_name  the master binlog file name
   @param pos       the master binlog file pos (the next after the event)

*/
static void inline slave_rows_error_report(enum loglevel level, int ha_error,
                                           rpl_group_info *rgi, THD *thd,
                                           TABLE *table, const char * type,
                                           const char *log_name, my_off_t pos)
{
  const char *handler_error= (ha_error ? HA_ERR(ha_error) : NULL);
199
  char buff[MAX_SLAVE_ERRMSG];
200 201 202 203 204 205 206 207 208 209 210 211 212
  Relay_log_info const *rli= rgi->rli;
  buff[0]= 0;
  int errcode= thd->is_error() ? thd->get_stmt_da()->sql_errno() : 0;

  /*
    In parallel replication, deadlocks or other temporary errors can happen
    occasionally in normal operation, they will be handled correctly and
    automatically by re-trying the transactions. So do not pollute the error
    log with messages about them.
  */
  if (is_parallel_retry_error(rgi, errcode))
    return;

213
  aggregate_da_errors(buff, sizeof(buff), thd->get_stmt_da());
214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533

  if (ha_error != 0)
    rli->report(level, errcode, rgi->gtid_info(),
                "Could not execute %s event on table %s.%s;"
                "%s handler error %s; "
                "the event's master log %s, end_log_pos %llu",
                type, table->s->db.str, table->s->table_name.str,
                buff, handler_error == NULL ? "<unknown>" : handler_error,
                log_name, pos);
  else
    rli->report(level, errcode, rgi->gtid_info(),
                "Could not execute %s event on table %s.%s;"
                "%s the event's master log %s, end_log_pos %llu",
                type, table->s->db.str, table->s->table_name.str,
                buff, log_name, pos);
}
#endif

#if defined(HAVE_REPLICATION)
static void set_thd_db(THD *thd, Rpl_filter *rpl_filter,
                       const char *db, uint32 db_len)
{
  char lcase_db_buf[NAME_LEN +1];
  LEX_CSTRING new_db;
  new_db.length= db_len;
  if (lower_case_table_names == 1)
  {
    strmov(lcase_db_buf, db);
    my_casedn_str(system_charset_info, lcase_db_buf);
    new_db.str= lcase_db_buf;
  }
  else
    new_db.str= db;
  /* TODO WARNING this makes rewrite_db respect lower_case_table_names values
   * for more info look MDEV-17446 */
  new_db.str= rpl_filter->get_rewrite_db(new_db.str, &new_db.length);
  thd->set_db(&new_db);
}
#endif


#if defined(HAVE_REPLICATION)

inline int idempotent_error_code(int err_code)
{
  int ret= 0;

  switch (err_code)
  {
    case 0:
      ret= 1;
    break;
    /*
      The following list of "idempotent" errors
      means that an error from the list might happen
      because of idempotent (more than once)
      applying of a binlog file.
      Notice, that binlog has a  ddl operation its
      second applying may cause

      case HA_ERR_TABLE_DEF_CHANGED:
      case HA_ERR_CANNOT_ADD_FOREIGN:

      which are not included into to the list.

      Note that HA_ERR_RECORD_DELETED is not in the list since
      do_exec_row() should not return that error code.
    */
    case HA_ERR_RECORD_CHANGED:
    case HA_ERR_KEY_NOT_FOUND:
    case HA_ERR_END_OF_FILE:
    case HA_ERR_FOUND_DUPP_KEY:
    case HA_ERR_FOUND_DUPP_UNIQUE:
    case HA_ERR_FOREIGN_DUPLICATE_KEY:
    case HA_ERR_NO_REFERENCED_ROW:
    case HA_ERR_ROW_IS_REFERENCED:
      ret= 1;
    break;
    default:
      ret= 0;
    break;
  }
  return (ret);
}

/**
  Ignore error code specified on command line.
*/

inline int ignored_error_code(int err_code)
{
  if (use_slave_mask && bitmap_is_set(&slave_error_mask, err_code))
  {
    statistic_increment(slave_skipped_errors, LOCK_status);
    return 1;
  }
  return err_code == ER_SLAVE_IGNORED_TABLE;
}

/*
  This function converts an engine's error to a server error.
   
  If the thread does not have an error already reported, it tries to 
  define it by calling the engine's method print_error. However, if a 
  mapping is not found, it uses the ER_UNKNOWN_ERROR and prints out a 
  warning message.
*/ 
int convert_handler_error(int error, THD* thd, TABLE *table)
{
  uint actual_error= (thd->is_error() ? thd->get_stmt_da()->sql_errno() :
                           0);

  if (actual_error == 0)
  {
    table->file->print_error(error, MYF(0));
    actual_error= (thd->is_error() ? thd->get_stmt_da()->sql_errno() :
                        ER_UNKNOWN_ERROR);
    if (actual_error == ER_UNKNOWN_ERROR)
      if (global_system_variables.log_warnings)
        sql_print_warning("Unknown error detected %d in handler", error);
  }

  return (actual_error);
}

inline bool concurrency_error_code(int error)
{
  switch (error)
  {
  case ER_LOCK_WAIT_TIMEOUT:
  case ER_LOCK_DEADLOCK:
  case ER_XA_RBDEADLOCK:
    return TRUE;
  default: 
    return (FALSE);
  }
}

inline bool unexpected_error_code(int unexpected_error)
{
  switch (unexpected_error) 
  {
  case ER_NET_READ_ERROR:
  case ER_NET_ERROR_ON_WRITE:
  case ER_QUERY_INTERRUPTED:
  case ER_STATEMENT_TIMEOUT:
  case ER_CONNECTION_KILLED:
  case ER_SERVER_SHUTDOWN:
  case ER_NEW_ABORTING_CONNECTION:
    return(TRUE);
  default:
    return(FALSE);
  }
}

/*
  pretty_print_str()
*/

static void
pretty_print_str(String *packet, const char *str, int len)
{
  const char *end= str + len;
  packet->append(STRING_WITH_LEN("'"));
  while (str < end)
  {
    char c;
    switch ((c=*str++)) {
    case '\n': packet->append(STRING_WITH_LEN("\\n")); break;
    case '\r': packet->append(STRING_WITH_LEN("\\r")); break;
    case '\\': packet->append(STRING_WITH_LEN("\\\\")); break;
    case '\b': packet->append(STRING_WITH_LEN("\\b")); break;
    case '\t': packet->append(STRING_WITH_LEN("\\t")); break;
    case '\'': packet->append(STRING_WITH_LEN("\\'")); break;
    case 0   : packet->append(STRING_WITH_LEN("\\0")); break;
    default:
      packet->append(&c, 1);
      break;
    }
  }
  packet->append(STRING_WITH_LEN("'"));
}
#endif /* HAVE_REPLICATION */


#if defined(HAVE_REPLICATION)

/**
  Create a prefix for the temporary files that is to be used for
  load data file name for this master

  @param name	           Store prefix of name here
  @param connection_name   Connection name
 
  @return pointer to end of name

  @description
  We assume that FN_REFLEN is big enough to hold
  MAX_CONNECTION_NAME * MAX_FILENAME_MBWIDTH characters + 2 numbers +
  a short extension.

  The resulting file name has the following parts, each separated with a '-'
  - PREFIX_SQL_LOAD (SQL_LOAD-)
  - If a connection name is given (multi-master setup):
    - Add an extra '-' to mark that this is a multi-master file
    - connection name in lower case, converted to safe file characters.
    (see create_logfile_name_with_suffix()).
  - server_id
  - A last '-' (after server_id).
*/

static char *load_data_tmp_prefix(char *name,
                                  LEX_CSTRING *connection_name)
{
  name= strmov(name, PREFIX_SQL_LOAD);
  if (connection_name->length)
  {
    uint buf_length;
    uint errors;
    /* Add marker that this is a multi-master-file */
    *name++='-';
    /* Convert connection_name to a safe filename */
    buf_length= strconvert(system_charset_info, connection_name->str, FN_REFLEN,
                           &my_charset_filename, name, FN_REFLEN, &errors);
    name+= buf_length;
    *name++= '-';
  }
  name= int10_to_str(global_system_variables.server_id, name, 10);
  *name++ = '-';
  *name= '\0';                                  // For testing prefixes
  return name;
}


/**
  Creates a temporary name for LOAD DATA INFILE

  @param buf		      Store new filename here
  @param file_id	      File_id (part of file name)
  @param event_server_id      Event_id (part of file name)
  @param ext		      Extension for file name

  @return
    Pointer to start of extension
*/

static char *slave_load_file_stem(char *buf, uint file_id,
                                  int event_server_id, const char *ext,
                                  LEX_CSTRING *connection_name)
{
  char *res;
  res= buf+ unpack_dirname(buf, slave_load_tmpdir);
  to_unix_path(buf);
  buf= load_data_tmp_prefix(res, connection_name);
  buf= int10_to_str(event_server_id, buf, 10);
  *buf++ = '-';
  res= int10_to_str(file_id, buf, 10);
  strmov(res, ext);                             // Add extension last
  return res;                                   // Pointer to extension
}
#endif


#if defined(HAVE_REPLICATION)

/**
  Delete all temporary files used for SQL_LOAD.
*/

static void cleanup_load_tmpdir(LEX_CSTRING *connection_name)
{
  MY_DIR *dirp;
  FILEINFO *file;
  uint i;
  char dir[FN_REFLEN], fname[FN_REFLEN];
  char prefbuf[31 + MAX_CONNECTION_NAME* MAX_FILENAME_MBWIDTH + 1];
  DBUG_ENTER("cleanup_load_tmpdir");

  unpack_dirname(dir, slave_load_tmpdir);
  if (!(dirp=my_dir(dir, MYF(MY_WME))))
    return;

  /* 
     When we are deleting temporary files, we should only remove
     the files associated with the server id of our server.
     We don't use event_server_id here because since we've disabled
     direct binlogging of Create_file/Append_file/Exec_load events
     we cannot meet Start_log event in the middle of events from one 
     LOAD DATA.
  */

  load_data_tmp_prefix(prefbuf, connection_name);
  DBUG_PRINT("enter", ("dir: '%s'  prefix: '%s'", dir, prefbuf));

  for (i=0 ; i < (uint)dirp->number_of_files; i++)
  {
    file=dirp->dir_entry+i;
    if (is_prefix(file->name, prefbuf))
    {
      fn_format(fname,file->name,slave_load_tmpdir,"",MY_UNPACK_FILENAME);
      mysql_file_delete(key_file_misc, fname, MYF(0));
    }
  }

  my_dirend(dirp);
  DBUG_VOID_RETURN;
}
#endif


/**
  Append a version of the 'str' string suitable for use in a query to
  the 'to' string.  To generate a correct escaping, the character set
  information in 'csinfo' is used.
*/

int append_query_string(CHARSET_INFO *csinfo, String *to,
                        const char *str, size_t len, bool no_backslash)
{
  char *beg, *ptr;
Monty's avatar
Monty committed
534
  my_bool overflow;
535 536 537 538 539 540 541 542 543 544 545 546 547
  uint32 const orig_len= to->length();
  if (to->reserve(orig_len + len * 2 + 4))
    return 1;

  beg= (char*) to->ptr() + to->length();
  ptr= beg;
  if (csinfo->escape_with_backslash_is_dangerous)
    ptr= str_to_hex(ptr, str, len);
  else
  {
    *ptr++= '\'';
    if (!no_backslash)
    {
Monty's avatar
Monty committed
548
      ptr+= escape_string_for_mysql(csinfo, ptr, 0, str, len, &overflow);
549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
    }
    else
    {
      const char *frm_str= str;

      for (; frm_str < (str + len); frm_str++)
      {
        /* Using '' way to represent "'" */
        if (*frm_str == '\'')
          *ptr++= *frm_str;

        *ptr++= *frm_str;
      }
    }

    *ptr++= '\'';
  }
  to->length((uint32)(orig_len + ptr - beg));
  return 0;
}


/**************************************************************************
	Log_event methods (= the parent class of all events)
**************************************************************************/

Log_event::Log_event(THD* thd_arg, uint16 flags_arg, bool using_trans)
  :log_pos(0), temp_buf(0), exec_time(0), thd(thd_arg),
   checksum_alg(BINLOG_CHECKSUM_ALG_UNDEF)
{
  server_id=	thd->variables.server_id;
  when=         thd->start_time;
  when_sec_part=thd->start_time_sec_part;

  if (using_trans)
    cache_type= Log_event::EVENT_TRANSACTIONAL_CACHE;
  else
    cache_type= Log_event::EVENT_STMT_CACHE;
  flags= flags_arg |
    (thd->variables.option_bits & OPTION_SKIP_REPLICATION ?
     LOG_EVENT_SKIP_REPLICATION_F : 0);
}

/**
  This minimal constructor is for when you are not even sure that there
  is a valid THD. For example in the server when we are shutting down or
  flushing logs after receiving a SIGHUP (then we must write a Rotate to
  the binlog but we have no THD, so we need this minimal constructor).
*/

Log_event::Log_event()
  :temp_buf(0), exec_time(0), flags(0), cache_type(EVENT_INVALID_CACHE),
   thd(0), checksum_alg(BINLOG_CHECKSUM_ALG_UNDEF)
{
  server_id=	global_system_variables.server_id;
  /*
    We can't call my_time() here as this would cause a call before
    my_init() is called
  */
  when=         0;
  when_sec_part=0;
  log_pos=	0;
}



#ifdef HAVE_REPLICATION

int Log_event::do_update_pos(rpl_group_info *rgi)
{
  Relay_log_info *rli= rgi->rli;
  DBUG_ENTER("Log_event::do_update_pos");

  DBUG_ASSERT(!rli->belongs_to_client());
  /*
    rli is null when (as far as I (Guilhem) know) the caller is
    Load_log_event::do_apply_event *and* that one is called from
    Execute_load_log_event::do_apply_event.  In this case, we don't
    do anything here ; Execute_load_log_event::do_apply_event will
    call Log_event::do_apply_event again later with the proper rli.
    Strictly speaking, if we were sure that rli is null only in the
    case discussed above, 'if (rli)' is useless here.  But as we are
    not 100% sure, keep it for now.

    Matz: I don't think we will need this check with this refactoring.
  */
  if (rli)
  {
    /*
      In parallel execution, delay position update for the events that are
      not part of event groups (format description, rotate, and such) until
      the actual event execution reaches that point.
    */
    if (!rgi->is_parallel_exec || is_group_event(get_type_code()))
      rli->stmt_done(log_pos, thd, rgi);
  }
  DBUG_RETURN(0);                                  // Cannot fail currently
}


Log_event::enum_skip_reason
Log_event::do_shall_skip(rpl_group_info *rgi)
{
  Relay_log_info *rli= rgi->rli;
  DBUG_PRINT("info", ("ev->server_id: %lu, ::server_id: %lu,"
                      " rli->replicate_same_server_id: %d,"
                      " rli->slave_skip_counter: %llu",
                      (ulong) server_id,
                      (ulong) global_system_variables.server_id,
                      rli->replicate_same_server_id,
                      rli->slave_skip_counter));
  if ((server_id == global_system_variables.server_id &&
661
       !(rli->replicate_same_server_id || (flags &  LOG_EVENT_ACCEPT_OWN_F))) ||
662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
      (rli->slave_skip_counter == 1 && rli->is_in_group()) ||
      (flags & LOG_EVENT_SKIP_REPLICATION_F &&
       opt_replicate_events_marked_for_skip != RPL_SKIP_REPLICATE))
    return EVENT_SKIP_IGNORE;
  if (rli->slave_skip_counter > 0)
    return EVENT_SKIP_COUNT;
  return EVENT_SKIP_NOT;
}


/*
  Log_event::pack_info()
*/

void Log_event::pack_info(Protocol *protocol)
{
Monty's avatar
Monty committed
678
  protocol->store("", 0, &my_charset_bin);
679 680 681 682 683 684 685 686 687 688 689 690 691 692
}


/**
  Only called by SHOW BINLOG EVENTS
*/
int Log_event::net_send(Protocol *protocol, const char* log_name, my_off_t pos)
{
  const char *p= strrchr(log_name, FN_LIBCHAR);
  const char *event_type;
  if (p)
    log_name = p + 1;

  protocol->prepare_for_resend();
Monty's avatar
Monty committed
693
  protocol->store(log_name, strlen(log_name), &my_charset_bin);
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753
  protocol->store((ulonglong) pos);
  event_type = get_type_str();
  protocol->store(event_type, strlen(event_type), &my_charset_bin);
  protocol->store((uint32) server_id);
  protocol->store((ulonglong) log_pos);
  pack_info(protocol);
  return protocol->write();
}
#endif /* HAVE_REPLICATION */


/**
  init_show_field_list() prepares the column names and types for the
  output of SHOW BINLOG EVENTS; it is used only by SHOW BINLOG
  EVENTS.
*/

void Log_event::init_show_field_list(THD *thd, List<Item>* field_list)
{
  MEM_ROOT *mem_root= thd->mem_root;
  field_list->push_back(new (mem_root)
                        Item_empty_string(thd, "Log_name", 20),
                        mem_root);
  field_list->push_back(new (mem_root)
                        Item_return_int(thd, "Pos",
                                        MY_INT64_NUM_DECIMAL_DIGITS,
                                        MYSQL_TYPE_LONGLONG),
                        mem_root);
  field_list->push_back(new (mem_root)
                        Item_empty_string(thd, "Event_type", 20),
                        mem_root);
  field_list->push_back(new (mem_root)
                        Item_return_int(thd, "Server_id", 10,
                                        MYSQL_TYPE_LONG),
                        mem_root);
  field_list->push_back(new (mem_root)
                        Item_return_int(thd, "End_log_pos",
                                        MY_INT64_NUM_DECIMAL_DIGITS,
                                        MYSQL_TYPE_LONGLONG),
                        mem_root);
  field_list->push_back(new (mem_root) Item_empty_string(thd, "Info", 20),
                        mem_root);
}

/**
   A decider of whether to trigger checksum computation or not.
   To be invoked in Log_event::write() stack.
   The decision is positive 

    S,M) if it's been marked for checksumming with @c checksum_alg
    
    M) otherwise, if @@global.binlog_checksum is not NONE and the event is 
       directly written to the binlog file.
       The to-be-cached event decides at @c write_cache() time.

   Otherwise the decision is negative.

   @note   A side effect of the method is altering Log_event::checksum_alg
           it the latter was undefined at calling.

Monty's avatar
Monty committed
754 755
   @return true   Checksum should be used. Log_event::checksum_alg is set.
   @return false  No checksum
756
*/
Monty's avatar
Monty committed
757

758 759 760
my_bool Log_event::need_checksum()
{
  my_bool ret;
Monty's avatar
Monty committed
761
  DBUG_ENTER("Log_event::need_checksum");
762

763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818
  /* 
     few callers of Log_event::write 
     (incl FD::write, FD constructing code on the slave side, Rotate relay log
     and Stop event) 
     provides their checksum alg preference through Log_event::checksum_alg.
  */
  if (checksum_alg != BINLOG_CHECKSUM_ALG_UNDEF)
    ret= checksum_alg != BINLOG_CHECKSUM_ALG_OFF;
  else
  {
    ret= binlog_checksum_options && cache_type == Log_event::EVENT_NO_CACHE;
    checksum_alg= ret ? (enum_binlog_checksum_alg)binlog_checksum_options
                      : BINLOG_CHECKSUM_ALG_OFF;
  }
  /*
    FD calls the methods before data_written has been calculated.
    The following invariant claims if the current is not the first
    call (and therefore data_written is not zero) then `ret' must be
    TRUE. It may not be null because FD is always checksummed.
  */
  
  DBUG_ASSERT(get_type_code() != FORMAT_DESCRIPTION_EVENT || ret ||
              data_written == 0);

  DBUG_ASSERT(!ret || 
              ((checksum_alg == binlog_checksum_options ||
               /* 
                  Stop event closes the relay-log and its checksum alg
                  preference is set by the caller can be different
                  from the server's binlog_checksum_options.
               */
               get_type_code() == STOP_EVENT ||
               /* 
                  Rotate:s can be checksummed regardless of the server's
                  binlog_checksum_options. That applies to both
                  the local RL's Rotate and the master's Rotate
                  which IO thread instantiates via queue_binlog_ver_3_event.
               */
               get_type_code() == ROTATE_EVENT ||
               get_type_code() == START_ENCRYPTION_EVENT ||
               /* FD is always checksummed */
               get_type_code() == FORMAT_DESCRIPTION_EVENT) && 
               checksum_alg != BINLOG_CHECKSUM_ALG_OFF));

  DBUG_ASSERT(checksum_alg != BINLOG_CHECKSUM_ALG_UNDEF);

  DBUG_ASSERT(((get_type_code() != ROTATE_EVENT &&
                get_type_code() != STOP_EVENT) ||
               get_type_code() != FORMAT_DESCRIPTION_EVENT) ||
              cache_type == Log_event::EVENT_NO_CACHE);

  DBUG_RETURN(ret);
}

int Log_event_writer::write_internal(const uchar *pos, size_t len)
{
819
  DBUG_ASSERT(!ctx || encrypt_or_write == &Log_event_writer::encrypt_and_write);
820
  if (my_b_safe_write(file, pos, len))
821 822
  {
    DBUG_PRINT("error", ("write to log failed: %d", my_errno));
823
    return 1;
824
  }
825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847
  bytes_written+= len;
  return 0;
}

/*
  as soon as encryption produces the first output block, write event_len
  where it should be in a valid event header
*/
int Log_event_writer::maybe_write_event_len(uchar *pos, size_t len)
{
  if (len && event_len)
  {
    DBUG_ASSERT(len >= EVENT_LEN_OFFSET);
    if (write_internal(pos + EVENT_LEN_OFFSET - 4, 4))
      return 1;
    int4store(pos + EVENT_LEN_OFFSET - 4, event_len);
    event_len= 0;
  }
  return 0;
}

int Log_event_writer::encrypt_and_write(const uchar *pos, size_t len)
{
848 849 850 851 852 853 854 855 856 857 858 859 860
  uchar *dst;
  size_t dstsize;
  uint dstlen;
  int res;                                      // Safe as res is always set
  DBUG_ASSERT(ctx);

  if (!len)
    return 0;

  dstsize= encryption_encrypted_length((uint)len, ENCRYPTION_KEY_SYSTEM_DATA,
                                       crypto->key_version);
  if (!(dst= (uchar*)my_safe_alloca(dstsize)))
    return 1;
861

862
  if (encryption_ctx_update(ctx, pos, (uint)len, dst, &dstlen))
863
  {
864 865
    res= 1;
    goto err;
866
  }
867 868 869 870

  if (maybe_write_event_len(dst, dstlen))
  {
    res= 1;
871
    goto err;
872 873 874
  }

  res= write_internal(dst, dstlen);
875 876 877

err:
  my_safe_afree(dst, dstsize);
878
  return res;
879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913
}

int Log_event_writer::write_header(uchar *pos, size_t len)
{
  DBUG_ENTER("Log_event_writer::write_header");
  /*
    recording checksum of FD event computed with dropped
    possibly active LOG_EVENT_BINLOG_IN_USE_F flag.
    Similar step at verication: the active flag is dropped before
    checksum computing.
  */
  if (checksum_len)
  {
    uchar save=pos[FLAGS_OFFSET];
    pos[FLAGS_OFFSET]&= ~LOG_EVENT_BINLOG_IN_USE_F;
    crc= my_checksum(0, pos, len);
    pos[FLAGS_OFFSET]= save;
  }

  if (ctx)
  {
    uchar iv[BINLOG_IV_LENGTH];
    crypto->set_iv(iv, (uint32)my_b_safe_tell(file));
    if (encryption_ctx_init(ctx, crypto->key, crypto->key_length,
           iv, sizeof(iv), ENCRYPTION_FLAG_ENCRYPT | ENCRYPTION_FLAG_NOPAD,
           ENCRYPTION_KEY_SYSTEM_DATA, crypto->key_version))
      DBUG_RETURN(1);

    DBUG_ASSERT(len >= LOG_EVENT_HEADER_LEN);
    event_len= uint4korr(pos + EVENT_LEN_OFFSET);
    DBUG_ASSERT(event_len >= len);
    memcpy(pos + EVENT_LEN_OFFSET, pos, 4);
    pos+= 4;
    len-= 4;
  }
914
  DBUG_RETURN((this->*encrypt_or_write)(pos, len));
915 916 917 918 919
}

int Log_event_writer::write_data(const uchar *pos, size_t len)
{
  DBUG_ENTER("Log_event_writer::write_data");
920 921 922 923

  if (!len)
    DBUG_RETURN(0);

924 925 926
  if (checksum_len)
    crc= my_checksum(crc, pos, len);

927
  DBUG_RETURN((this->*encrypt_or_write)(pos, len));
928 929 930 931 932 933 934 935 936
}

int Log_event_writer::write_footer()
{
  DBUG_ENTER("Log_event_writer::write_footer");
  if (checksum_len)
  {
    uchar checksum_buf[BINLOG_CHECKSUM_LEN];
    int4store(checksum_buf, crc);
937
    if ((this->*encrypt_or_write)(checksum_buf, BINLOG_CHECKSUM_LEN))
938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
      DBUG_RETURN(ER_ERROR_ON_WRITE);
  }
  if (ctx)
  {
    uint dstlen;
    uchar dst[MY_AES_BLOCK_SIZE*2];
    if (encryption_ctx_finish(ctx, dst, &dstlen))
      DBUG_RETURN(1);
    if (maybe_write_event_len(dst, dstlen) || write_internal(dst, dstlen))
      DBUG_RETURN(ER_ERROR_ON_WRITE);
  }
  DBUG_RETURN(0);
}

/*
  Log_event::write_header()
*/

bool Log_event::write_header(size_t event_data_length)
{
  uchar header[LOG_EVENT_HEADER_LEN];
  ulong now;
  DBUG_ENTER("Log_event::write_header");
  DBUG_PRINT("enter", ("filepos: %lld  length: %zu type: %d",
                       (longlong) writer->pos(), event_data_length,
                       (int) get_type_code()));

  writer->checksum_len= need_checksum() ? BINLOG_CHECKSUM_LEN : 0;

  /* Store number of bytes that will be written by this event */
  data_written= event_data_length + sizeof(header) + writer->checksum_len;

  /*
    log_pos != 0 if this is relay-log event. In this case we should not
    change the position
  */

  if (is_artificial_event())
  {
    /*
      Artificial events are automatically generated and do not exist
      in master's binary log, so log_pos should be set to 0.
    */
    log_pos= 0;
  }
  else  if (!log_pos)
  {
    /*
      Calculate the position of where the next event will start
      (end of this event, that is).
    */

    log_pos= writer->pos() + data_written;
    
    DBUG_EXECUTE_IF("dbug_master_binlog_over_2GB", log_pos += (1ULL <<31););
  }

  now= get_time();                               // Query start time

  /*
    Header will be of size LOG_EVENT_HEADER_LEN for all events, except for
    FORMAT_DESCRIPTION_EVENT and ROTATE_EVENT, where it will be
    LOG_EVENT_MINIMAL_HEADER_LEN (remember these 2 have a frozen header,
    because we read them before knowing the format).
  */

  int4store(header, now);              // timestamp
  header[EVENT_TYPE_OFFSET]= get_type_code();
  int4store(header+ SERVER_ID_OFFSET, server_id);
  int4store(header+ EVENT_LEN_OFFSET, data_written);
  int4store(header+ LOG_POS_OFFSET, log_pos);
  int2store(header + FLAGS_OFFSET, flags);

  bool ret= writer->write_header(header, sizeof(header));
  DBUG_RETURN(ret);
}



#if defined(HAVE_REPLICATION)
inline Log_event::enum_skip_reason
Log_event::continue_group(rpl_group_info *rgi)
{
  if (rgi->rli->slave_skip_counter == 1)
    return Log_event::EVENT_SKIP_IGNORE;
  return Log_event::do_shall_skip(rgi);
}
#endif

/**************************************************************************
	Query_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)

/**
  This (which is used only for SHOW BINLOG EVENTS) could be updated to
  print SET @@session_var=. But this is not urgent, as SHOW BINLOG EVENTS is
  only an information, it does not produce suitable queries to replay (for
  example it does not print LOAD DATA INFILE).
  @todo
    show the catalog ??
*/

void Query_log_event::pack_info(Protocol *protocol)
{
  // TODO: show the catalog ??
  char buf_mem[1024];
  String buf(buf_mem, sizeof(buf_mem), system_charset_info);
  buf.real_alloc(9 + db_len + q_len);
  if (!(flags & LOG_EVENT_SUPPRESS_USE_F)
      && db && db_len)
  {
    buf.append(STRING_WITH_LEN("use "));
    append_identifier(protocol->thd, &buf, db, db_len);
    buf.append(STRING_WITH_LEN("; "));
  }
1055 1056 1057

  DBUG_ASSERT(!flags2 || flags2_inited);

1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
  if (flags2 & (OPTION_NO_FOREIGN_KEY_CHECKS | OPTION_AUTO_IS_NULL |
                OPTION_RELAXED_UNIQUE_CHECKS |
                OPTION_NO_CHECK_CONSTRAINT_CHECKS |
                OPTION_IF_EXISTS))
  {
    buf.append(STRING_WITH_LEN("set "));
    if (flags2 & OPTION_NO_FOREIGN_KEY_CHECKS)
      buf.append(STRING_WITH_LEN("foreign_key_checks=1, "));
    if (flags2 & OPTION_AUTO_IS_NULL)
      buf.append(STRING_WITH_LEN("sql_auto_is_null, "));
    if (flags2 & OPTION_RELAXED_UNIQUE_CHECKS)
      buf.append(STRING_WITH_LEN("unique_checks=1, "));
    if (flags2 & OPTION_NO_CHECK_CONSTRAINT_CHECKS)
      buf.append(STRING_WITH_LEN("check_constraint_checks=1, "));
    if (flags2 & OPTION_IF_EXISTS)
      buf.append(STRING_WITH_LEN("@@sql_if_exists=1, "));
    buf[buf.length()-2]=';';
  }
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
  if (query && q_len)
    buf.append(query, q_len);
  protocol->store(&buf);
}
#endif


/**
  Utility function for the next method (Query_log_event::write()) .
*/
static void store_str_with_code_and_len(uchar **dst, const char *src,
                                        uint len, uint code)
{
  /*
    only 1 byte to store the length of catalog, so it should not
    surpass 255
  */
  DBUG_ASSERT(len <= 255);
  DBUG_ASSERT(src);
  *((*dst)++)= (uchar) code;
  *((*dst)++)= (uchar) len;
  bmove(*dst, src, len);
  (*dst)+= len;
}


/**
  Query_log_event::write().

  @note
    In this event we have to modify the header to have the correct
    EVENT_LEN_OFFSET as we don't yet know how many status variables we
    will print!
*/

bool Query_log_event::write()
{
  uchar buf[QUERY_HEADER_LEN + MAX_SIZE_LOG_EVENT_STATUS];
  uchar *start, *start_of_status;
  ulong event_length;

  if (!query)
    return 1;                                   // Something wrong with event

  /*
    We want to store the thread id:
    (- as an information for the user when he reads the binlog)
    - if the query uses temporary table: for the slave SQL thread to know to
    which master connection the temp table belongs.
    Now imagine we (write()) are called by the slave SQL thread (we are
    logging a query executed by this thread; the slave runs with
    --log-slave-updates). Then this query will be logged with
    thread_id=the_thread_id_of_the_SQL_thread. Imagine that 2 temp tables of
    the same name were created simultaneously on the master (in the master
    binlog you have
    CREATE TEMPORARY TABLE t; (thread 1)
    CREATE TEMPORARY TABLE t; (thread 2)
    ...)
    then in the slave's binlog there will be
    CREATE TEMPORARY TABLE t; (thread_id_of_the_slave_SQL_thread)
    CREATE TEMPORARY TABLE t; (thread_id_of_the_slave_SQL_thread)
    which is bad (same thread id!).

    To avoid this, we log the thread's thread id EXCEPT for the SQL
    slave thread for which we log the original (master's) thread id.
    Now this moves the bug: what happens if the thread id on the
    master was 10 and when the slave replicates the query, a
    connection number 10 is opened by a normal client on the slave,
    and updates a temp table of the same name? We get a problem
    again. To avoid this, in the handling of temp tables (sql_base.cc)
    we use thread_id AND server_id.  TODO when this is merged into
    4.1: in 4.1, slave_proxy_id has been renamed to pseudo_thread_id
    and is a session variable: that's to make mysqlbinlog work with
    temp tables. We probably need to introduce

    SET PSEUDO_SERVER_ID
    for mysqlbinlog in 4.1. mysqlbinlog would print:
    SET PSEUDO_SERVER_ID=
    SET PSEUDO_THREAD_ID=
    for each query using temp tables.
  */
  int4store(buf + Q_THREAD_ID_OFFSET, slave_proxy_id);
  int4store(buf + Q_EXEC_TIME_OFFSET, exec_time);
  buf[Q_DB_LEN_OFFSET] = (char) db_len;
  int2store(buf + Q_ERR_CODE_OFFSET, error_code);

  /*
    You MUST always write status vars in increasing order of code. This
    guarantees that a slightly older slave will be able to parse those he
    knows.
  */
  start_of_status= start= buf+QUERY_HEADER_LEN;
  if (flags2_inited)
  {
    *start++= Q_FLAGS2_CODE;
    int4store(start, flags2);
    start+= 4;
  }
  if (sql_mode_inited)
  {
    *start++= Q_SQL_MODE_CODE;
    int8store(start, (ulonglong)sql_mode);
    start+= 8;
  }
  if (catalog_len) // i.e. this var is inited (false for 4.0 events)
  {
    store_str_with_code_and_len(&start,
                                catalog, catalog_len, Q_CATALOG_NZ_CODE);
    /*
      In 5.0.x where x<4 masters we used to store the end zero here. This was
      a waste of one byte so we don't do it in x>=4 masters. We change code to
      Q_CATALOG_NZ_CODE, because re-using the old code would make x<4 slaves
      of this x>=4 master segfault (expecting a zero when there is
      none). Remaining compatibility problems are: the older slave will not
      find the catalog; but it is will not crash, and it's not an issue
      that it does not find the catalog as catalogs were not used in these
      older MySQL versions (we store it in binlog and read it from relay log
      but do nothing useful with it). What is an issue is that the older slave
      will stop processing the Q_* blocks (and jumps to the db/query) as soon
      as it sees unknown Q_CATALOG_NZ_CODE; so it will not be able to read
      Q_AUTO_INCREMENT*, Q_CHARSET and so replication will fail silently in
      various ways. Documented that you should not mix alpha/beta versions if
      they are not exactly the same version, with example of 5.0.3->5.0.2 and
      5.0.4->5.0.3. If replication is from older to new, the new will
      recognize Q_CATALOG_CODE and have no problem.
    */
  }
  if (auto_increment_increment != 1 || auto_increment_offset != 1)
  {
    *start++= Q_AUTO_INCREMENT;
    int2store(start, auto_increment_increment);
    int2store(start+2, auto_increment_offset);
    start+= 4;
  }
  if (charset_inited)
  {
    *start++= Q_CHARSET_CODE;
    memcpy(start, charset, 6);
    start+= 6;
  }
  if (time_zone_len)
  {
    /* In the TZ sys table, column Name is of length 64 so this should be ok */
    DBUG_ASSERT(time_zone_len <= MAX_TIME_ZONE_NAME_LENGTH);
    store_str_with_code_and_len(&start,
                                time_zone_str, time_zone_len, Q_TIME_ZONE_CODE);
  }
  if (lc_time_names_number)
  {
    DBUG_ASSERT(lc_time_names_number <= 0xFFFF);
    *start++= Q_LC_TIME_NAMES_CODE;
    int2store(start, lc_time_names_number);
    start+= 2;
  }
  if (charset_database_number)
  {
    DBUG_ASSERT(charset_database_number <= 0xFFFF);
    *start++= Q_CHARSET_DATABASE_CODE;
    int2store(start, charset_database_number);
    start+= 2;
  }
  if (table_map_for_update)
  {
    *start++= Q_TABLE_MAP_FOR_UPDATE_CODE;
    int8store(start, table_map_for_update);
    start+= 8;
  }
  if (master_data_written != 0)
  {
    /*
      Q_MASTER_DATA_WRITTEN_CODE only exists in relay logs where the master
      has binlog_version<4 and the slave has binlog_version=4. See comment
      for master_data_written in log_event.h for details.
    */
    *start++= Q_MASTER_DATA_WRITTEN_CODE;
    int4store(start, master_data_written);
    start+= 4;
  }

  if (thd && thd->need_binlog_invoker())
  {
    LEX_CSTRING user;
    LEX_CSTRING host;
    memset(&user, 0, sizeof(user));
    memset(&host, 0, sizeof(host));

    if (thd->slave_thread && thd->has_invoker())
    {
      /* user will be null, if master is older than this patch */
      user= thd->get_invoker_user();
      host= thd->get_invoker_host();
    }
    else
    {
      Security_context *ctx= thd->security_ctx;

      if (thd->need_binlog_invoker() == THD::INVOKER_USER)
      {
        user.str= ctx->priv_user;
        host.str= ctx->priv_host;
        host.length= strlen(host.str);
      }
      else
      {
        user.str= ctx->priv_role;
        host= empty_clex_str;
      }
      user.length= strlen(user.str);
    }

    if (user.length > 0)
    {
      *start++= Q_INVOKER;

      /*
        Store user length and user. The max length of use is 16, so 1 byte is
        enough to store the user's length.
       */
      *start++= (uchar)user.length;
      memcpy(start, user.str, user.length);
      start+= user.length;

      /*
        Store host length and host. The max length of host is 60, so 1 byte is
        enough to store the host's length.
       */
      *start++= (uchar)host.length;
      memcpy(start, host.str, host.length);
      start+= host.length;
    }
  }

  if (thd && thd->query_start_sec_part_used)
  {
    *start++= Q_HRNOW;
    get_time();
    int3store(start, when_sec_part);
    start+= 3;
  }
Monty's avatar
Monty committed
1315 1316 1317 1318 1319

  /* xid's is used with ddl_log handling */
  if (thd && thd->binlog_xid)
  {
    *start++= Q_XID;
1320
    int8store(start, thd->binlog_xid);
Monty's avatar
Monty committed
1321 1322 1323
    start+= 8;
  }

1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
  /*
    NOTE: When adding new status vars, please don't forget to update
    the MAX_SIZE_LOG_EVENT_STATUS in log_event.h and update the function
    code_name() in this file.
   
    Here there could be code like
    if (command-line-option-which-says-"log_this_variable" && inited)
    {
    *start++= Q_THIS_VARIABLE_CODE;
    int4store(start, this_variable);
    start+= 4;
    }
  */
  
  /* Store length of status variables */
  status_vars_len= (uint) (start-start_of_status);
  DBUG_ASSERT(status_vars_len <= MAX_SIZE_LOG_EVENT_STATUS);
  int2store(buf + Q_STATUS_VARS_LEN_OFFSET, status_vars_len);

  /*
    Calculate length of whole event
    The "1" below is the \0 in the db's length
  */
Monty's avatar
Monty committed
1347 1348
  event_length= ((uint) (start-buf) + get_post_header_size_for_derived() +
                 db_len + 1 + q_len);
1349 1350 1351 1352

  return write_header(event_length) ||
         write_data(buf, QUERY_HEADER_LEN) ||
         write_post_header_for_derived() ||
Monty's avatar
Monty committed
1353 1354
         write_data(start_of_status, (uint) status_vars_len) ||
         write_data(db, db_len + 1) ||
1355 1356 1357 1358 1359 1360
         write_data(query, q_len) ||
         write_footer();
}

bool Query_compressed_log_event::write()
{
1361
  uchar *buffer;
Monty's avatar
Monty committed
1362 1363 1364 1365
  uint32 alloc_size, compressed_size;
  bool ret= true;

  compressed_size= alloc_size= binlog_get_compress_len(q_len);
1366
  buffer= (uchar*) my_safe_alloca(alloc_size);
Monty's avatar
Monty committed
1367
  if (buffer &&
1368
      !binlog_buf_compress((uchar*) query, buffer, q_len, &compressed_size))
1369
  {
Monty's avatar
Monty committed
1370 1371 1372 1373 1374 1375
    /*
      Write the compressed event. We have to temporarily store the event
      in query and q_len as Query_log_event::write() uses these.
    */
    const char *query_tmp= query;
    uint32 q_len_tmp= q_len;
1376
    query= (char*) buffer;
Monty's avatar
Monty committed
1377 1378 1379 1380
    q_len= compressed_size;
    ret= Query_log_event::write();
    query= query_tmp;
    q_len= q_len_tmp;
1381
  }
Monty's avatar
Monty committed
1382
  my_safe_afree(buffer, alloc_size);
1383 1384 1385
  return ret;
}

Monty's avatar
Monty committed
1386

1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414
/**
  The simplest constructor that could possibly work.  This is used for
  creating static objects that have a special meaning and are invisible
  to the log.  
*/
Query_log_event::Query_log_event()
  :Log_event(), data_buf(0)
{
  memset(&user, 0, sizeof(user));
  memset(&host, 0, sizeof(host));
}


/*
  SYNOPSIS
    Query_log_event::Query_log_event()
      thd_arg           - thread handle
      query_arg         - array of char representing the query
      query_length      - size of the  `query_arg' array
      using_trans       - there is a modified transactional table
      direct            - Don't cache statement
      suppress_use      - suppress the generation of 'USE' statements
      errcode           - the error code of the query
      
  DESCRIPTION
  Creates an event for binlogging
  The value for `errcode' should be supplied by caller.
*/
1415 1416
Query_log_event::Query_log_event(THD* thd_arg, const char* query_arg,
                                 size_t query_length, bool using_trans,
1417 1418 1419 1420 1421 1422 1423 1424
				 bool direct, bool suppress_use, int errcode)

  :Log_event(thd_arg,
             (thd_arg->thread_specific_used ? LOG_EVENT_THREAD_SPECIFIC_F :
              0) |
             (suppress_use ? LOG_EVENT_SUPPRESS_USE_F : 0),
	     using_trans),
   data_buf(0), query(query_arg), catalog(thd_arg->catalog),
Monty's avatar
Monty committed
1425
   q_len((uint32) query_length),
1426 1427 1428
   thread_id(thd_arg->thread_id),
   /* save the original thread id; we already know the server id */
   slave_proxy_id((ulong)thd_arg->variables.pseudo_thread_id),
1429
   flags2_inited(1), sql_mode_inited(1), charset_inited(1), flags2(0),
1430 1431 1432 1433 1434 1435 1436 1437
   sql_mode(thd_arg->variables.sql_mode),
   auto_increment_increment(thd_arg->variables.auto_increment_increment),
   auto_increment_offset(thd_arg->variables.auto_increment_offset),
   lc_time_names_number(thd_arg->variables.lc_time_names->number),
   charset_database_number(0),
   table_map_for_update((ulonglong)thd_arg->table_map_for_update),
   master_data_written(0)
{
Monty's avatar
Monty committed
1438 1439
  /* status_vars_len is set just before writing the event */

1440 1441 1442 1443 1444 1445
  time_t end_time;

#ifdef WITH_WSREP
  /*
    If Query_log_event will contain non trans keyword (not BEGIN, COMMIT,
    SAVEPOINT or ROLLBACK) we disable PA for this transaction.
Marko Mäkelä's avatar
Marko Mäkelä committed
1446 1447 1448
    Note that here WSREP(thd) might not be true e.g. when wsrep_shcema
    is created we create tables with thd->variables.wsrep_on=false
    to avoid replicating wsrep_schema tables to other nodes.
1449 1450
   */
  if (WSREP_ON && !is_trans_keyword())
Marko Mäkelä's avatar
Marko Mäkelä committed
1451
  {
1452
    thd->wsrep_PA_safe= false;
Marko Mäkelä's avatar
Marko Mäkelä committed
1453
  }
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466
#endif /* WITH_WSREP */

  memset(&user, 0, sizeof(user));
  memset(&host, 0, sizeof(host));
  error_code= errcode;

  end_time= my_time(0);
  exec_time = (ulong) (end_time  - thd_arg->start_time);
  /**
    @todo this means that if we have no catalog, then it is replicated
    as an existing catalog of length zero. is that safe? /sven
  */
  catalog_len = (catalog) ? (uint32) strlen(catalog) : 0;
Monty's avatar
Monty committed
1467 1468 1469 1470

  if (!(db= thd->db.str))
    db= "";
  db_len= (uint32) strlen(db);
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
  if (thd_arg->variables.collation_database != thd_arg->db_charset)
    charset_database_number= thd_arg->variables.collation_database->number;
  
  /*
    We only replicate over the bits of flags2 that we need: the rest
    are masked out by "& OPTIONS_WRITTEN_TO_BINLOG".

    We also force AUTOCOMMIT=1.  Rationale (cf. BUG#29288): After
    fixing BUG#26395, we always write BEGIN and COMMIT around all
    transactions (even single statements in autocommit mode).  This is
    so that replication from non-transactional to transactional table
    and error recovery from XA to non-XA table should work as
    expected.  The BEGIN/COMMIT are added in log.cc. However, there is
    one exception: MyISAM bypasses log.cc and writes directly to the
    binlog.  So if autocommit is off, master has MyISAM, and slave has
    a transactional engine, then the slave will just see one long
    never-ending transaction.  The only way to bypass explicit
    BEGIN/COMMIT in the binlog is by using a non-transactional table.
    So setting AUTOCOMMIT=1 will make this work as expected.

    Note: explicitly replicate AUTOCOMMIT=1 from master. We do not
    assume AUTOCOMMIT=1 on slave; the slave still reads the state of
    the autocommit flag as written by the master to the binlog. This
    behavior may change after WL#4162 has been implemented.
  */
  flags2= (uint32) (thd_arg->variables.option_bits &
                    (OPTIONS_WRITTEN_TO_BIN_LOG & ~OPTION_NOT_AUTOCOMMIT));
  DBUG_ASSERT(thd_arg->variables.character_set_client->number < 256*256);
  DBUG_ASSERT(thd_arg->variables.collation_connection->number < 256*256);
  DBUG_ASSERT(thd_arg->variables.collation_server->number < 256*256);
  DBUG_ASSERT(thd_arg->variables.character_set_client->mbminlen == 1);
  int2store(charset, thd_arg->variables.character_set_client->number);
  int2store(charset+2, thd_arg->variables.collation_connection->number);
  int2store(charset+4, thd_arg->variables.collation_server->number);
  if (thd_arg->time_zone_used)
  {
    /*
      Note that our event becomes dependent on the Time_zone object
      representing the time zone. Fortunately such objects are never deleted
      or changed during mysqld's lifetime.
    */
    time_zone_len= thd_arg->variables.time_zone->get_name()->length();
    time_zone_str= thd_arg->variables.time_zone->get_name()->ptr();
  }
  else
    time_zone_len= 0;

  LEX *lex= thd->lex;
  /*
    Defines that the statement will be written directly to the binary log
    without being wrapped by a BEGIN...COMMIT. Otherwise, the statement
    will be written to either the trx-cache or stmt-cache.

    Note that a cache will not be used if the parameter direct is TRUE.
  */
  bool use_cache= FALSE;
  /*
    TRUE defines that the trx-cache must be used and by consequence the
    use_cache is TRUE.

    Note that a cache will not be used if the parameter direct is TRUE.
  */
  bool trx_cache= FALSE;
  cache_type= Log_event::EVENT_INVALID_CACHE;

Marko Mäkelä's avatar
Marko Mäkelä committed
1536
  if (!direct)
1537
  {
Marko Mäkelä's avatar
Marko Mäkelä committed
1538 1539 1540 1541 1542 1543
    switch (lex->sql_command)
    {
      case SQLCOM_DROP_TABLE:
      case SQLCOM_DROP_SEQUENCE:
        use_cache= (lex->tmp_table() && thd->in_multi_stmt_transaction_mode());
        break;
1544

Marko Mäkelä's avatar
Marko Mäkelä committed
1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
      case SQLCOM_CREATE_TABLE:
      case SQLCOM_CREATE_SEQUENCE:
        /*
           If we are using CREATE ... SELECT or if we are a slave
           executing BEGIN...COMMIT (generated by CREATE...SELECT) we
           have to use the transactional cache to ensure we don't
           calculate any checksum for the CREATE part.
         */
        trx_cache= (lex->first_select_lex()->item_list.elements &&
            thd->is_current_stmt_binlog_format_row()) ||
          (thd->variables.option_bits & OPTION_GTID_BEGIN);
        use_cache= (lex->tmp_table() &&
            thd->in_multi_stmt_transaction_mode()) || trx_cache;
        break;
      case SQLCOM_SET_OPTION:
        if (lex->autocommit)
          use_cache= trx_cache= FALSE;
        else
          use_cache= TRUE;
        break;
      case SQLCOM_RELEASE_SAVEPOINT:
      case SQLCOM_ROLLBACK_TO_SAVEPOINT:
      case SQLCOM_SAVEPOINT:
1568
      case SQLCOM_XA_END:
Marko Mäkelä's avatar
Marko Mäkelä committed
1569 1570 1571 1572 1573 1574
        use_cache= trx_cache= TRUE;
        break;
      default:
        use_cache= sqlcom_can_generate_row_events(thd);
        break;
    }
1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
  }

  if (!use_cache || direct)
  {
    cache_type= Log_event::EVENT_NO_CACHE;
  }
  else if (using_trans || trx_cache || stmt_has_updated_trans_table(thd) ||
           thd->lex->is_mixed_stmt_unsafe(thd->in_multi_stmt_transaction_mode(),
                                          thd->variables.binlog_direct_non_trans_update,
                                          trans_has_updated_trans_table(thd),
                                          thd->tx_isolation))
    cache_type= Log_event::EVENT_TRANSACTIONAL_CACHE;
  else
    cache_type= Log_event::EVENT_STMT_CACHE;
  DBUG_ASSERT(cache_type != Log_event::EVENT_INVALID_CACHE);
  DBUG_PRINT("info",("Query_log_event has flags2: %lu  sql_mode: %llu  cache_tye: %d",
                     (ulong) flags2, sql_mode, cache_type));
}

Query_compressed_log_event::Query_compressed_log_event(THD* thd_arg, const char* query_arg,
    ulong query_length, bool using_trans,
    bool direct, bool suppress_use, int errcode)
    :Query_log_event(thd_arg, query_arg, query_length, using_trans, direct,
                     suppress_use, errcode),
     query_buf(0)
{

}


#if defined(HAVE_REPLICATION)

int Query_log_event::do_apply_event(rpl_group_info *rgi)
{
  return do_apply_event(rgi, query, q_len);
}

/**
   Compare if two errors should be regarded as equal.
   This is to handle the case when you can get slightly different errors
   on master and slave for the same thing.
   @param
   expected_error	Error we got on master
   actual_error		Error we got on slave

   @return
   1 Errors are equal
   0 Errors are different
*/

bool test_if_equal_repl_errors(int expected_error, int actual_error)
{
  if (expected_error == actual_error)
    return 1;
  switch (expected_error) {
  case ER_DUP_ENTRY:
  case ER_DUP_ENTRY_WITH_KEY_NAME:
  case ER_DUP_KEY:
  case ER_AUTOINC_READ_FAILED:
    return (actual_error == ER_DUP_ENTRY ||
            actual_error == ER_DUP_ENTRY_WITH_KEY_NAME ||
            actual_error == ER_DUP_KEY ||
            actual_error == ER_AUTOINC_READ_FAILED ||
            actual_error == HA_ERR_AUTOINC_ERANGE);
  case ER_UNKNOWN_TABLE:
    return actual_error == ER_IT_IS_A_VIEW;
  default:
    break;
  }
  return 0;
}


/**
  @todo
  Compare the values of "affected rows" around here. Something
  like:
  @code
     if ((uint32) affected_in_event != (uint32) affected_on_slave)
     {
     sql_print_error("Slave: did not get the expected number of affected \
     rows running query from master - expected %d, got %d (this numbers \
     should have matched modulo 4294967296).", 0, ...);
     thd->query_error = 1;
     }
  @endcode
  We may also want an option to tell the slave to ignore "affected"
  mismatch. This mismatch could be implemented with a new ER_ code, and
  to ignore it you would use --slave-skip-errors...
*/
int Query_log_event::do_apply_event(rpl_group_info *rgi,
                                    const char *query_arg, uint32 q_len_arg)
{
  int expected_error,actual_error= 0;
  Schema_specification_st db_options;
  uint64 sub_id= 0;
  void *hton= NULL;
  rpl_gtid gtid;
  Relay_log_info const *rli= rgi->rli;
  Rpl_filter *rpl_filter= rli->mi->rpl_filter;
  bool current_stmt_is_commit;
  DBUG_ENTER("Query_log_event::do_apply_event");

  /*
    Colleagues: please never free(thd->catalog) in MySQL. This would
    lead to bugs as here thd->catalog is a part of an alloced block,
    not an entire alloced block (see
    Query_log_event::do_apply_event()). Same for thd->db.  Thank
    you.
  */
  thd->catalog= catalog_len ? (char *) catalog : (char *)"";

  size_t valid_len= Well_formed_prefix(system_charset_info,
                                       db, db_len, NAME_LEN).length();

  if (valid_len != db_len)
  {
    rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                "Invalid database name in Query event.");
    thd->is_slave_error= true;
    goto end;
  }

  set_thd_db(thd, rpl_filter, db, db_len);

  /*
    Setting the character set and collation of the current database thd->db.
   */
  load_db_opt_by_name(thd, thd->db.str, &db_options);
  if (db_options.default_table_charset)
    thd->db_charset= db_options.default_table_charset;
  thd->variables.auto_increment_increment= auto_increment_increment;
  thd->variables.auto_increment_offset=    auto_increment_offset;

  DBUG_PRINT("info", ("log_pos: %lu", (ulong) log_pos));

  thd->clear_error(1);
  current_stmt_is_commit= is_commit();

  DBUG_ASSERT(!current_stmt_is_commit || !rgi->tables_to_lock);
  rgi->slave_close_thread_tables(thd);

  /*
    Note:   We do not need to execute reset_one_shot_variables() if this
            db_ok() test fails.
    Reason: The db stored in binlog events is the same for SET and for
            its companion query.  If the SET is ignored because of
            db_ok(), the companion query will also be ignored, and if
            the companion query is ignored in the db_ok() test of
            ::do_apply_event(), then the companion SET also have so
            we don't need to reset_one_shot_variables().
  */
  if (is_trans_keyword() || rpl_filter->db_ok(thd->db.str))
  {
    thd->set_time(when, when_sec_part);
    thd->set_query_and_id((char*)query_arg, q_len_arg,
                          thd->charset(), next_query_id());
    thd->variables.pseudo_thread_id= thread_id;		// for temp tables
    DBUG_PRINT("query",("%s", thd->query()));

    if (unlikely(!(expected_error= error_code)) ||
        ignored_error_code(expected_error) ||
        !unexpected_error_code(expected_error))
    {
      thd->slave_expected_error= expected_error;
      if (flags2_inited)
1742
      {
1743
        ulonglong mask= flags2_inited;
1744 1745
        thd->variables.option_bits= (flags2 & mask) |
                                    (thd->variables.option_bits & ~mask);
1746
      }
1747 1748
      /*
        else, we are in a 3.23/4.0 binlog; we previously received a
1749 1750
        Rotate_log_event which reset thd->variables.option_bits and
        sql_mode etc, so nothing to do.
1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
      */
      /*
        We do not replicate MODE_NO_DIR_IN_CREATE. That is, if the master is a
        slave which runs with SQL_MODE=MODE_NO_DIR_IN_CREATE, this should not
        force us to ignore the dir too. Imagine you are a ring of machines, and
        one has a disk problem so that you temporarily need
        MODE_NO_DIR_IN_CREATE on this machine; you don't want it to propagate
        elsewhere (you don't want all slaves to start ignoring the dirs).
      */
      if (sql_mode_inited)
        thd->variables.sql_mode=
          (sql_mode_t) ((thd->variables.sql_mode & MODE_NO_DIR_IN_CREATE) |
                        (sql_mode & ~(sql_mode_t) MODE_NO_DIR_IN_CREATE));
      if (charset_inited)
      {
        rpl_sql_thread_info *sql_info= thd->system_thread_info.rpl_sql_info;
        if (sql_info->cached_charset_compare(charset))
        {
          /* Verify that we support the charsets found in the event. */
          if (!(thd->variables.character_set_client=
                get_charset(uint2korr(charset), MYF(MY_WME))) ||
              !(thd->variables.collation_connection=
                get_charset(uint2korr(charset+2), MYF(MY_WME))) ||
              !(thd->variables.collation_server=
                get_charset(uint2korr(charset+4), MYF(MY_WME))))
          {
            /*
              We updated the thd->variables with nonsensical values (0). Let's
              set them to something safe (i.e. which avoids crash), and we'll
              stop with EE_UNKNOWN_CHARSET in compare_errors (unless set to
              ignore this error).
            */
            set_slave_thread_default_charset(thd, rgi);
            goto compare_errors;
          }
          thd->update_charset(); // for the charset change to take effect
          /*
            Reset thd->query_string.cs to the newly set value.
            Note, there is a small flaw here. For a very short time frame
            if the new charset is different from the old charset and
            if another thread executes "SHOW PROCESSLIST" after
            the above thd->set_query_and_id() and before this thd->set_query(),
            and if the current query has some non-ASCII characters,
            the another thread may see some '?' marks in the PROCESSLIST
            result. This should be acceptable now. This is a reminder
            to fix this if any refactoring happens here sometime.
          */
          thd->set_query((char*) query_arg, q_len_arg, thd->charset());
        }
      }
      if (time_zone_len)
      {
        String tmp(time_zone_str, time_zone_len, &my_charset_bin);
        if (!(thd->variables.time_zone= my_tz_find(thd, &tmp)))
        {
          my_error(ER_UNKNOWN_TIME_ZONE, MYF(0), tmp.c_ptr());
          thd->variables.time_zone= global_system_variables.time_zone;
          goto compare_errors;
        }
      }
      if (lc_time_names_number)
      {
        if (!(thd->variables.lc_time_names=
              my_locale_by_number(lc_time_names_number)))
        {
          my_printf_error(ER_UNKNOWN_ERROR,
                      "Unknown locale: '%d'", MYF(0), lc_time_names_number);
          thd->variables.lc_time_names= &my_locale_en_US;
          goto compare_errors;
        }
      }
      else
        thd->variables.lc_time_names= &my_locale_en_US;
      if (charset_database_number)
      {
        CHARSET_INFO *cs;
        if (!(cs= get_charset(charset_database_number, MYF(0))))
        {
          char buf[20];
          int10_to_str((int) charset_database_number, buf, -10);
          my_error(ER_UNKNOWN_COLLATION, MYF(0), buf);
          goto compare_errors;
        }
        thd->variables.collation_database= cs;
      }
      else
        thd->variables.collation_database= thd->db_charset;

      {
        const CHARSET_INFO *cs= thd->charset();
        /*
          We cannot ask for parsing a statement using a character set
          without state_maps (parser internal data).
        */
        if (!cs->state_map)
        {
          rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                      ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                      "character_set cannot be parsed");
          thd->is_slave_error= true;
          goto end;
        }
      }

      /*
        Record any GTID in the same transaction, so slave state is
        transactionally consistent.
      */
      if (current_stmt_is_commit)
      {
        thd->variables.option_bits&= ~OPTION_GTID_BEGIN;
        if (rgi->gtid_pending)
        {
          sub_id= rgi->gtid_sub_id;
          rgi->gtid_pending= false;

          gtid= rgi->current_gtid;
          if (unlikely(rpl_global_gtid_slave_state->record_gtid(thd, &gtid,
                                                                sub_id,
                                                                true, false,
                                                                &hton)))
          {
            int errcode= thd->get_stmt_da()->sql_errno();
            if (!is_parallel_retry_error(rgi, errcode))
              rli->report(ERROR_LEVEL, ER_CANNOT_UPDATE_GTID_STATE,
                          rgi->gtid_info(),
                          "Error during COMMIT: failed to update GTID state in "
                        "%s.%s: %d: %s",
                          "mysql", rpl_gtid_slave_state_table_name.str,
                          errcode,
                          thd->get_stmt_da()->message());
            sub_id= 0;
            thd->is_slave_error= 1;
            goto end;
          }
        }
      }

      thd->table_map_for_update= (table_map)table_map_for_update;
      thd->set_invoker(&user, &host);
      /*
        Flag if we need to rollback the statement transaction on
        slave if it by chance succeeds.
        If we expected a non-zero error code and get nothing and,
        it is a concurrency issue or ignorable issue, effects
        of the statement should be rolled back.
      */
      if (unlikely(expected_error) &&
          (ignored_error_code(expected_error) ||
           concurrency_error_code(expected_error)))
      {
        thd->variables.option_bits|= OPTION_MASTER_SQL_ERROR;
        thd->variables.option_bits&= ~OPTION_GTID_BEGIN;
      }
      /* Execute the query (note that we bypass dispatch_command()) */
      Parser_state parser_state;
      if (!parser_state.init(thd, thd->query(), thd->query_length()))
      {
        DBUG_ASSERT(thd->m_digest == NULL);
        thd->m_digest= & thd->m_digest_state;
        DBUG_ASSERT(thd->m_statement_psi == NULL);
        thd->m_statement_psi= MYSQL_START_STATEMENT(&thd->m_statement_state,
                                                    stmt_info_rpl.m_key,
                                                    thd->db.str, thd->db.length,
1915 1916
                                                    thd->charset(), NULL);
        THD_STAGE_INFO(thd, stage_starting);
1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
        MYSQL_SET_STATEMENT_TEXT(thd->m_statement_psi, thd->query(), thd->query_length());
        if (thd->m_digest != NULL)
          thd->m_digest->reset(thd->m_token_array, max_digest_length);

         if (thd->slave_thread)
         {
           /*
             To be compatible with previous releases, the slave thread uses the global
             log_slow_disabled_statements value, wich can be changed dynamically, so we
             have to set the sql_log_slow respectively.
           */
           thd->variables.sql_log_slow= !MY_TEST(global_system_variables.log_slow_disabled_statements & LOG_SLOW_DISABLE_SLAVE);
         }

1931
        mysql_parse(thd, thd->query(), thd->query_length(), &parser_state);
1932 1933 1934 1935
        /* Finalize server status flags after executing a statement. */
        thd->update_server_status();
        log_slow_statement(thd);
        thd->lex->restore_set_statement_var();
Marko Mäkelä's avatar
Marko Mäkelä committed
1936 1937 1938 1939 1940 1941 1942

        /*
          When THD::slave_expected_error gets reset inside execution stack
          that is the case of to be ignored event. In this case the expected
          error must change to the reset value as well.
        */
        expected_error= thd->slave_expected_error;
1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056
      }

      thd->variables.option_bits&= ~OPTION_MASTER_SQL_ERROR;
    }
    else
    {
      /*
        The query got a really bad error on the master (thread killed etc),
        which could be inconsistent. Parse it to test the table names: if the
        replicate-*-do|ignore-table rules say "this query must be ignored" then
        we exit gracefully; otherwise we warn about the bad error and tell DBA
        to check/fix it.
      */
      if (mysql_test_parse_for_slave(thd, thd->query(), thd->query_length()))
        thd->clear_error(1);
      else
      {
        rli->report(ERROR_LEVEL, expected_error, rgi->gtid_info(),
                          "\
Query partially completed on the master (error on master: %d) \
and was aborted. There is a chance that your master is inconsistent at this \
point. If you are sure that your master is ok, run this query manually on the \
slave and then restart the slave with SET GLOBAL SQL_SLAVE_SKIP_COUNTER=1; \
START SLAVE; . Query: '%s'", expected_error, thd->query());
        thd->is_slave_error= 1;
      }
      goto end;
    }

    /* If the query was not ignored, it is printed to the general log */
    if (likely(!thd->is_error()) ||
        thd->get_stmt_da()->sql_errno() != ER_SLAVE_IGNORED_TABLE)
      general_log_write(thd, COM_QUERY, thd->query(), thd->query_length());
    else
    {
      /*
        Bug#54201: If we skip an INSERT query that uses auto_increment, then we
        should reset any @@INSERT_ID set by an Intvar_log_event associated with
        the query; otherwise the @@INSERT_ID will linger until the next INSERT
        that uses auto_increment and may affect extra triggers on the slave etc.

        We reset INSERT_ID unconditionally; it is probably cheaper than
        checking if it is necessary.
      */
      thd->auto_inc_intervals_forced.empty();
    }

compare_errors:
    /*
      In the slave thread, we may sometimes execute some DROP / * 40005
      TEMPORARY * / TABLE that come from parts of binlogs (likely if we
      use RESET SLAVE or CHANGE MASTER TO), while the temporary table
      has already been dropped. To ignore such irrelevant "table does
      not exist errors", we silently clear the error if TEMPORARY was used.
    */
    if ((thd->lex->sql_command == SQLCOM_DROP_TABLE ||
         thd->lex->sql_command == SQLCOM_DROP_SEQUENCE) &&
        thd->lex->tmp_table() &&
        thd->is_error() && thd->get_stmt_da()->sql_errno() == ER_BAD_TABLE_ERROR &&
        !expected_error)
      thd->get_stmt_da()->reset_diagnostics_area();
    /*
      If we expected a non-zero error code, and we don't get the same error
      code, and it should be ignored or is related to a concurrency issue.
    */
    actual_error= thd->is_error() ? thd->get_stmt_da()->sql_errno() : 0;
    DBUG_PRINT("info",("expected_error: %d  sql_errno: %d",
                       expected_error, actual_error));

    if ((unlikely(expected_error) &&
         !test_if_equal_repl_errors(expected_error, actual_error) &&
         !concurrency_error_code(expected_error)) &&
        !ignored_error_code(actual_error) &&
        !ignored_error_code(expected_error))
    {
      rli->report(ERROR_LEVEL, 0, rgi->gtid_info(),
                  "Query caused different errors on master and slave.     "
                  "Error on master: message (format)='%s' error code=%d ; "
                  "Error on slave: actual message='%s', error code=%d. "
                  "Default database: '%s'. Query: '%s'",
                  ER_THD(thd, expected_error),
                  expected_error,
                  actual_error ? thd->get_stmt_da()->message() : "no error",
                  actual_error,
                  print_slave_db_safe(db), query_arg);
      thd->is_slave_error= 1;
    }
    /*
      If we get the same error code as expected and it is not a concurrency
      issue, or should be ignored.
    */
    else if ((test_if_equal_repl_errors(expected_error, actual_error) &&
              !concurrency_error_code(expected_error)) ||
             ignored_error_code(actual_error))
    {
      DBUG_PRINT("info",("error ignored"));
      thd->clear_error(1);
      if (actual_error == ER_QUERY_INTERRUPTED ||
          actual_error == ER_CONNECTION_KILLED)
        thd->reset_killed();
    }
    /*
      Other cases: mostly we expected no error and get one.
    */
    else if (unlikely(thd->is_slave_error || thd->is_fatal_error))
    {
      if (!is_parallel_retry_error(rgi, actual_error))
        rli->report(ERROR_LEVEL, actual_error, rgi->gtid_info(),
                    "Error '%s' on query. Default database: '%s'. Query: '%s'",
                    (actual_error ? thd->get_stmt_da()->message() :
                     "unexpected success or fatal error"),
                    thd->get_db(), query_arg);
      thd->is_slave_error= 1;
#ifdef WITH_WSREP
Marko Mäkelä's avatar
Marko Mäkelä committed
2057
      if (wsrep_thd_is_toi(thd) && wsrep_must_ignore_error(thd))
2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098
      {
        thd->clear_error(1);
        thd->killed= NOT_KILLED;
        thd->wsrep_has_ignored_error= true;
      }
#endif /* WITH_WSREP */
    }

    /*
      TODO: compare the values of "affected rows" around here. Something
      like:
      if ((uint32) affected_in_event != (uint32) affected_on_slave)
      {
      sql_print_error("Slave: did not get the expected number of affected \
      rows running query from master - expected %d, got %d (this numbers \
      should have matched modulo 4294967296).", 0, ...);
      thd->is_slave_error = 1;
      }
      We may also want an option to tell the slave to ignore "affected"
      mismatch. This mismatch could be implemented with a new ER_ code, and
      to ignore it you would use --slave-skip-errors...

      To do the comparison we need to know the value of "affected" which the
      above mysql_parse() computed. And we need to know the value of
      "affected" in the master's binlog. Both will be implemented later. The
      important thing is that we now have the format ready to log the values
      of "affected" in the binlog. So we can release 5.0.0 before effectively
      logging "affected" and effectively comparing it.
    */
  } /* End of if (db_ok(... */

  {
    /**
      The following failure injecion works in cooperation with tests
      setting @@global.debug= 'd,stop_slave_middle_group'.
      The sql thread receives the killed status and will proceed
      to shutdown trying to finish incomplete events group.
    */
    DBUG_EXECUTE_IF("stop_slave_middle_group",
                    if (!current_stmt_is_commit && is_begin() == 0)
                    {
2099
                      if (thd->transaction->all.modified_non_trans_table)
2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173
                        const_cast<Relay_log_info*>(rli)->abort_slave= 1;
                    };);
  }

end:
  if (unlikely(sub_id && !thd->is_slave_error))
    rpl_global_gtid_slave_state->update_state_hash(sub_id, &gtid, hton, rgi);

  /*
    Probably we have set thd->query, thd->db, thd->catalog to point to places
    in the data_buf of this event. Now the event is going to be deleted
    probably, so data_buf will be freed, so the thd->... listed above will be
    pointers to freed memory.
    So we must set them to 0, so that those bad pointers values are not later
    used. Note that "cleanup" queries like automatic DROP TEMPORARY TABLE
    don't suffer from these assignments to 0 as DROP TEMPORARY
    TABLE uses the db.table syntax.
  */
  thd->catalog= 0;
  thd->set_db(&null_clex_str);    /* will free the current database */
  thd->reset_query();
  DBUG_PRINT("info", ("end: query= 0"));

  /* Mark the statement completed. */
  MYSQL_END_STATEMENT(thd->m_statement_psi, thd->get_stmt_da());
  thd->m_statement_psi= NULL;
  thd->m_digest= NULL;

  /*
    As a disk space optimization, future masters will not log an event for
    LAST_INSERT_ID() if that function returned 0 (and thus they will be able
    to replace the THD::stmt_depends_on_first_successful_insert_id_in_prev_stmt
    variable by (THD->first_successful_insert_id_in_prev_stmt > 0) ; with the
    resetting below we are ready to support that.
  */
  thd->first_successful_insert_id_in_prev_stmt_for_binlog= 0;
  thd->first_successful_insert_id_in_prev_stmt= 0;
  thd->stmt_depends_on_first_successful_insert_id_in_prev_stmt= 0;
  free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));
  DBUG_RETURN(thd->is_slave_error);
}

Log_event::enum_skip_reason
Query_log_event::do_shall_skip(rpl_group_info *rgi)
{
  Relay_log_info *rli= rgi->rli;
  DBUG_ENTER("Query_log_event::do_shall_skip");
  DBUG_PRINT("debug", ("query: '%s'  q_len: %d", query, q_len));
  DBUG_ASSERT(query && q_len > 0);
  DBUG_ASSERT(thd == rgi->thd);

  /*
    An event skipped due to @@skip_replication must not be counted towards the
    number of events to be skipped due to @@sql_slave_skip_counter.
  */
  if (flags & LOG_EVENT_SKIP_REPLICATION_F &&
      opt_replicate_events_marked_for_skip != RPL_SKIP_REPLICATE)
    DBUG_RETURN(Log_event::EVENT_SKIP_IGNORE);

  if (rli->slave_skip_counter > 0)
  {
    if (is_begin())
    {
      thd->variables.option_bits|= OPTION_BEGIN | OPTION_GTID_BEGIN;
      DBUG_RETURN(Log_event::continue_group(rgi));
    }

    if (is_commit() || is_rollback())
    {
      thd->variables.option_bits&= ~(OPTION_BEGIN | OPTION_GTID_BEGIN);
      DBUG_RETURN(Log_event::EVENT_SKIP_COUNT);
    }
  }
#ifdef WITH_WSREP
Marko Mäkelä's avatar
Marko Mäkelä committed
2174 2175
  else if (WSREP(thd) && wsrep_mysql_replication_bundle &&
           opt_slave_domain_parallel_threads == 0 &&
2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
           thd->wsrep_mysql_replicated > 0 &&
           (is_begin() || is_commit()))
  {
    if (++thd->wsrep_mysql_replicated < (int)wsrep_mysql_replication_bundle)
    {
      WSREP_DEBUG("skipping wsrep commit %d", thd->wsrep_mysql_replicated);
      DBUG_RETURN(Log_event::EVENT_SKIP_IGNORE);
    }
    else
    {
      thd->wsrep_mysql_replicated = 0;
    }
  }
Marko Mäkelä's avatar
Marko Mäkelä committed
2189
#endif /* WITH_WSREP */
2190 2191 2192 2193 2194
  DBUG_RETURN(Log_event::do_shall_skip(rgi));
}


bool
2195
Query_log_event::peek_is_commit_rollback(const uchar *event_start,
2196
                                         size_t event_len,
2197 2198
                                         enum enum_binlog_checksum_alg
                                         checksum_alg)
2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420
{
  if (checksum_alg == BINLOG_CHECKSUM_ALG_CRC32)
  {
    if (event_len > BINLOG_CHECKSUM_LEN)
      event_len-= BINLOG_CHECKSUM_LEN;
    else
      event_len= 0;
  }
  else
    DBUG_ASSERT(checksum_alg == BINLOG_CHECKSUM_ALG_UNDEF ||
                checksum_alg == BINLOG_CHECKSUM_ALG_OFF);

  if (event_len < LOG_EVENT_HEADER_LEN + QUERY_HEADER_LEN || event_len < 9)
    return false;
  return !memcmp(event_start + (event_len-7), "\0COMMIT", 7) ||
         !memcmp(event_start + (event_len-9), "\0ROLLBACK", 9);
}

#endif


/**************************************************************************
	Start_log_event_v3 methods
**************************************************************************/

Start_log_event_v3::Start_log_event_v3()
  :Log_event(), created(0), binlog_version(BINLOG_VERSION),
   dont_set_created(0)
{
  memcpy(server_version, ::server_version, ST_SERVER_VER_LEN);
}


#if defined(HAVE_REPLICATION)
void Start_log_event_v3::pack_info(Protocol *protocol)
{
  char buf[12 + ST_SERVER_VER_LEN + 14 + 22], *pos;
  pos= strmov(buf, "Server ver: ");
  pos= strmov(pos, server_version);
  pos= strmov(pos, ", Binlog ver: ");
  pos= int10_to_str(binlog_version, pos, 10);
  protocol->store(buf, (uint) (pos-buf), &my_charset_bin);
}
#endif


bool Start_log_event_v3::write()
{
  char buff[START_V3_HEADER_LEN];
  int2store(buff + ST_BINLOG_VER_OFFSET,binlog_version);
  memcpy(buff + ST_SERVER_VER_OFFSET,server_version,ST_SERVER_VER_LEN);
  if (!dont_set_created)
    created= get_time(); // this sets when and when_sec_part as a side effect
  int4store(buff + ST_CREATED_OFFSET,created);
  return write_header(sizeof(buff)) ||
         write_data(buff, sizeof(buff)) ||
         write_footer();
}


#if defined(HAVE_REPLICATION)

/**
  Start_log_event_v3::do_apply_event() .
  The master started

    IMPLEMENTATION
    - To handle the case where the master died without having time to write
    DROP TEMPORARY TABLE, DO RELEASE_LOCK (prepared statements' deletion is
    TODO), we clean up all temporary tables that we got, if we are sure we
    can (see below).

  @todo
    - Remove all active user locks.
    Guilhem 2003-06: this is true but not urgent: the worst it can cause is
    the use of a bit of memory for a user lock which will not be used
    anymore. If the user lock is later used, the old one will be released. In
    other words, no deadlock problem.
*/

int Start_log_event_v3::do_apply_event(rpl_group_info *rgi)
{
  DBUG_ENTER("Start_log_event_v3::do_apply_event");
  int error= 0;
  Relay_log_info *rli= rgi->rli;

  switch (binlog_version)
  {
  case 3:
  case 4:
    /*
      This can either be 4.x (then a Start_log_event_v3 is only at master
      startup so we are sure the master has restarted and cleared his temp
      tables; the event always has 'created'>0) or 5.0 (then we have to test
      'created').
    */
    if (created)
    {
      rli->close_temporary_tables();
      
      /*
        The following is only false if we get here with a BINLOG statement
      */
      if (rli->mi)
        cleanup_load_tmpdir(&rli->mi->cmp_connection_name);
    }
    break;

    /*
       Now the older formats; in that case load_tmpdir is cleaned up by the I/O
       thread.
    */
  case 1:
    if (strncmp(rli->relay_log.description_event_for_exec->server_version,
                "3.23.57",7) >= 0 && created)
    {
      /*
        Can distinguish, based on the value of 'created': this event was
        generated at master startup.
      */
      rli->close_temporary_tables();
    }
    /*
      Otherwise, can't distinguish a Start_log_event generated at
      master startup and one generated by master FLUSH LOGS, so cannot
      be sure temp tables have to be dropped. So do nothing.
    */
    break;
  default:
    /*
      This case is not expected. It can be either an event corruption or an
      unsupported binary log version.
    */
    rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                "Binlog version not supported");
    DBUG_RETURN(1);
  }
  DBUG_RETURN(error);
}
#endif /* defined(HAVE_REPLICATION) */

/***************************************************************************
       Format_description_log_event methods
****************************************************************************/

bool Format_description_log_event::write()
{
  bool ret;
  bool no_checksum;
  /*
    We don't call Start_log_event_v3::write() because this would make 2
    my_b_safe_write().
  */
  uchar buff[START_V3_HEADER_LEN+1];
  size_t rec_size= sizeof(buff) + BINLOG_CHECKSUM_ALG_DESC_LEN +
                   number_of_event_types;
  int2store(buff + ST_BINLOG_VER_OFFSET,binlog_version);
  memcpy((char*) buff + ST_SERVER_VER_OFFSET,server_version,ST_SERVER_VER_LEN);
  if (!dont_set_created)
    created= get_time();
  int4store(buff + ST_CREATED_OFFSET,created);
  buff[ST_COMMON_HEADER_LEN_OFFSET]= common_header_len;
  /*
    if checksum is requested
    record the checksum-algorithm descriptor next to
    post_header_len vector which will be followed by the checksum value.
    Master is supposed to trigger checksum computing by binlog_checksum_options,
    slave does it via marking the event according to
    FD_queue checksum_alg value.
  */
  compile_time_assert(BINLOG_CHECKSUM_ALG_DESC_LEN == 1);
#ifdef DBUG_ASSERT_EXISTS
  data_written= 0; // to prepare for need_checksum assert
#endif
  uint8 checksum_byte= (uint8)
    (need_checksum() ? checksum_alg : BINLOG_CHECKSUM_ALG_OFF);
  /* 
     FD of checksum-aware server is always checksum-equipped, (V) is in,
     regardless of @@global.binlog_checksum policy.
     Thereby a combination of (A) == 0, (V) != 0 means
     it's the checksum-aware server's FD event that heads checksum-free binlog
     file. 
     Here 0 stands for checksumming OFF to evaluate (V) as 0 is that case.
     A combination of (A) != 0, (V) != 0 denotes FD of the checksum-aware server
     heading the checksummed binlog.
     (A), (V) presence in FD of the checksum-aware server makes the event
     1 + 4 bytes bigger comparing to the former FD.
  */

  if ((no_checksum= (checksum_alg == BINLOG_CHECKSUM_ALG_OFF)))
  {
    checksum_alg= BINLOG_CHECKSUM_ALG_CRC32;  // Forcing (V) room to fill anyway
  }
  ret= write_header(rec_size) ||
       write_data(buff, sizeof(buff)) ||
       write_data(post_header_len, number_of_event_types) ||
       write_data(&checksum_byte, sizeof(checksum_byte)) ||
       write_footer();
  if (no_checksum)
    checksum_alg= BINLOG_CHECKSUM_ALG_OFF;
  return ret;
}

#if defined(HAVE_REPLICATION)
int Format_description_log_event::do_apply_event(rpl_group_info *rgi)
{
  int ret= 0;
  Relay_log_info *rli= rgi->rli;
  DBUG_ENTER("Format_description_log_event::do_apply_event");

  /*
    As a transaction NEVER spans on 2 or more binlogs:
    if we have an active transaction at this point, the master died
    while writing the transaction to the binary log, i.e. while
    flushing the binlog cache to the binlog. XA guarantees that master has
    rolled back. So we roll back.
    Note: this event could be sent by the master to inform us of the
    format of its binlog; in other words maybe it is not at its
    original place when it comes to us; we'll know this by checking
    log_pos ("artificial" events have log_pos == 0).
  */
2421 2422
  if (!thd->rli_fake &&
      !is_artificial_event() && created && thd->transaction->all.ha_list)
2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768
  {
    /* This is not an error (XA is safe), just an information */
    rli->report(INFORMATION_LEVEL, 0, NULL,
                "Rolling back unfinished transaction (no COMMIT "
                "or ROLLBACK in relay log). A probable cause is that "
                "the master died while writing the transaction to "
                "its binary log, thus rolled back too."); 
    rgi->cleanup_context(thd, 1);
  }

  /*
    If this event comes from ourselves, there is no cleaning task to
    perform, we don't call Start_log_event_v3::do_apply_event()
    (this was just to update the log's description event).
  */
  if (server_id != (uint32) global_system_variables.server_id)
  {
    /*
      If the event was not requested by the slave i.e. the master sent
      it while the slave asked for a position >4, the event will make
      rli->group_master_log_pos advance. Say that the slave asked for
      position 1000, and the Format_desc event's end is 96. Then in
      the beginning of replication rli->group_master_log_pos will be
      0, then 96, then jump to first really asked event (which is
      >96). So this is ok.
    */
    ret= Start_log_event_v3::do_apply_event(rgi);
  }

  if (!ret)
  {
    /* Save the information describing this binlog */
    copy_crypto_data(rli->relay_log.description_event_for_exec);
    delete rli->relay_log.description_event_for_exec;
    rli->relay_log.description_event_for_exec= this;
  }

  DBUG_RETURN(ret);
}

int Format_description_log_event::do_update_pos(rpl_group_info *rgi)
{
  if (server_id == (uint32) global_system_variables.server_id)
  {
    /*
      We only increase the relay log position if we are skipping
      events and do not touch any group_* variables, nor flush the
      relay log info.  If there is a crash, we will have to re-skip
      the events again, but that is a minor issue.

      If we do not skip stepping the group log position (and the
      server id was changed when restarting the server), it might well
      be that we start executing at a position that is invalid, e.g.,
      at a Rows_log_event or a Query_log_event preceeded by a
      Intvar_log_event instead of starting at a Table_map_log_event or
      the Intvar_log_event respectively.
     */
    rgi->inc_event_relay_log_pos();
    return 0;
  }
  else
  {
    return Log_event::do_update_pos(rgi);
  }
}

Log_event::enum_skip_reason
Format_description_log_event::do_shall_skip(rpl_group_info *rgi)
{
  return Log_event::EVENT_SKIP_NOT;
}

#endif


#if defined(HAVE_REPLICATION)
int Start_encryption_log_event::do_apply_event(rpl_group_info* rgi)
{
  return rgi->rli->relay_log.description_event_for_exec->start_decryption(this);
}

int Start_encryption_log_event::do_update_pos(rpl_group_info *rgi)
{
  /*
    master never sends Start_encryption_log_event, any SELE that a slave
    might see was created locally in MYSQL_BIN_LOG::open() on the slave
  */
  rgi->inc_event_relay_log_pos();
  return 0;
}

#endif


/**************************************************************************
      Load_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
bool Load_log_event::print_query(THD *thd, bool need_db, const char *cs,
                                 String *buf, my_off_t *fn_start,
                                 my_off_t *fn_end, const char *qualify_db)
{
  if (need_db && db && db_len)
  {
    buf->append(STRING_WITH_LEN("use "));
    append_identifier(thd, buf, db, db_len);
    buf->append(STRING_WITH_LEN("; "));
  }

  buf->append(STRING_WITH_LEN("LOAD DATA "));

  if (is_concurrent)
    buf->append(STRING_WITH_LEN("CONCURRENT "));

  if (fn_start)
    *fn_start= buf->length();

  if (check_fname_outside_temp_buf())
    buf->append(STRING_WITH_LEN("LOCAL "));
  buf->append(STRING_WITH_LEN("INFILE '"));
  buf->append_for_single_quote(fname, fname_len);
  buf->append(STRING_WITH_LEN("' "));

  if (sql_ex.opt_flags & REPLACE_FLAG)
    buf->append(STRING_WITH_LEN("REPLACE "));
  else if (sql_ex.opt_flags & IGNORE_FLAG)
    buf->append(STRING_WITH_LEN("IGNORE "));

  buf->append(STRING_WITH_LEN("INTO"));

  if (fn_end)
    *fn_end= buf->length();

  buf->append(STRING_WITH_LEN(" TABLE "));
  if (qualify_db)
  {
    append_identifier(thd, buf, qualify_db, strlen(qualify_db));
    buf->append(STRING_WITH_LEN("."));
  }
  append_identifier(thd, buf, table_name, table_name_len);

  if (cs != NULL)
  {
    buf->append(STRING_WITH_LEN(" CHARACTER SET "));
    buf->append(cs, strlen(cs));
  }

  /* We have to create all optional fields as the default is not empty */
  buf->append(STRING_WITH_LEN(" FIELDS TERMINATED BY "));
  pretty_print_str(buf, sql_ex.field_term, sql_ex.field_term_len);
  if (sql_ex.opt_flags & OPT_ENCLOSED_FLAG)
    buf->append(STRING_WITH_LEN(" OPTIONALLY "));
  buf->append(STRING_WITH_LEN(" ENCLOSED BY "));
  pretty_print_str(buf, sql_ex.enclosed, sql_ex.enclosed_len);

  buf->append(STRING_WITH_LEN(" ESCAPED BY "));
  pretty_print_str(buf, sql_ex.escaped, sql_ex.escaped_len);

  buf->append(STRING_WITH_LEN(" LINES TERMINATED BY "));
  pretty_print_str(buf, sql_ex.line_term, sql_ex.line_term_len);
  if (sql_ex.line_start_len)
  {
    buf->append(STRING_WITH_LEN(" STARTING BY "));
    pretty_print_str(buf, sql_ex.line_start, sql_ex.line_start_len);
  }

  if ((long) skip_lines > 0)
  {
    buf->append(STRING_WITH_LEN(" IGNORE "));
    buf->append_ulonglong(skip_lines);
    buf->append(STRING_WITH_LEN(" LINES "));
  }

  if (num_fields)
  {
    uint i;
    const char *field= fields;
    buf->append(STRING_WITH_LEN(" ("));
    for (i = 0; i < num_fields; i++)
    {
      if (i)
      {
        /*
          Yes, the space and comma is reversed here. But this is mostly dead
          code, at most used when reading really old binlogs from old servers,
          so better just leave it as is...
        */
        buf->append(STRING_WITH_LEN(" ,"));
      }
      append_identifier(thd, buf, field, field_lens[i]);
      field+= field_lens[i]  + 1;
    }
    buf->append(STRING_WITH_LEN(")"));
  }
  return 0;
}


void Load_log_event::pack_info(Protocol *protocol)
{
  char query_buffer[1024];
  String query_str(query_buffer, sizeof(query_buffer), system_charset_info);

  query_str.length(0);
  print_query(protocol->thd, TRUE, NULL, &query_str, 0, 0, NULL);
  protocol->store(query_str.ptr(), query_str.length(), &my_charset_bin);
}
#endif /* defined(HAVE_REPLICATION) */


bool Load_log_event::write_data_header()
{
  char buf[LOAD_HEADER_LEN];
  int4store(buf + L_THREAD_ID_OFFSET, slave_proxy_id);
  int4store(buf + L_EXEC_TIME_OFFSET, exec_time);
  int4store(buf + L_SKIP_LINES_OFFSET, skip_lines);
  buf[L_TBL_LEN_OFFSET] = (char)table_name_len;
  buf[L_DB_LEN_OFFSET] = (char)db_len;
  int4store(buf + L_NUM_FIELDS_OFFSET, num_fields);
  return write_data(buf, LOAD_HEADER_LEN) != 0;
}


bool Load_log_event::write_data_body()
{
  if (sql_ex.write_data(writer))
    return 1;
  if (num_fields && fields && field_lens)
  {
    if (write_data(field_lens, num_fields) ||
	write_data(fields, field_block_len))
      return 1;
  }
  return (write_data(table_name, table_name_len + 1) ||
	  write_data(db, db_len + 1) ||
	  write_data(fname, fname_len));
}


Load_log_event::Load_log_event(THD *thd_arg, const sql_exchange *ex,
			       const char *db_arg, const char *table_name_arg,
			       List<Item> &fields_arg,
                               bool is_concurrent_arg,
			       enum enum_duplicates handle_dup,
			       bool ignore, bool using_trans)
  :Log_event(thd_arg,
             thd_arg->thread_specific_used ? LOG_EVENT_THREAD_SPECIFIC_F : 0,
             using_trans),
   thread_id(thd_arg->thread_id),
   slave_proxy_id((ulong)thd_arg->variables.pseudo_thread_id),
   num_fields(0),fields(0),
   field_lens(0),field_block_len(0),
   table_name(table_name_arg ? table_name_arg : ""),
   db(db_arg), fname(ex->file_name), local_fname(FALSE),
   is_concurrent(is_concurrent_arg)
{
  time_t end_time;
  time(&end_time);
  exec_time = (ulong) (end_time  - thd_arg->start_time);
  /* db can never be a zero pointer in 4.0 */
  db_len = (uint32) strlen(db);
  table_name_len = (uint32) strlen(table_name);
  fname_len = (fname) ? (uint) strlen(fname) : 0;
  sql_ex.field_term = ex->field_term->ptr();
  sql_ex.field_term_len = (uint8) ex->field_term->length();
  sql_ex.enclosed = ex->enclosed->ptr();
  sql_ex.enclosed_len = (uint8) ex->enclosed->length();
  sql_ex.line_term = ex->line_term->ptr();
  sql_ex.line_term_len = (uint8) ex->line_term->length();
  sql_ex.line_start = ex->line_start->ptr();
  sql_ex.line_start_len = (uint8) ex->line_start->length();
  sql_ex.escaped = ex->escaped->ptr();
  sql_ex.escaped_len = (uint8) ex->escaped->length();
  sql_ex.opt_flags = 0;
  sql_ex.cached_new_format = -1;
    
  if (ex->dumpfile)
    sql_ex.opt_flags|= DUMPFILE_FLAG;
  if (ex->opt_enclosed)
    sql_ex.opt_flags|= OPT_ENCLOSED_FLAG;

  sql_ex.empty_flags= 0;

  switch (handle_dup) {
  case DUP_REPLACE:
    sql_ex.opt_flags|= REPLACE_FLAG;
    break;
  case DUP_UPDATE:				// Impossible here
  case DUP_ERROR:
    break;	
  }
  if (ignore)
    sql_ex.opt_flags|= IGNORE_FLAG;

  if (!ex->field_term->length())
    sql_ex.empty_flags |= FIELD_TERM_EMPTY;
  if (!ex->enclosed->length())
    sql_ex.empty_flags |= ENCLOSED_EMPTY;
  if (!ex->line_term->length())
    sql_ex.empty_flags |= LINE_TERM_EMPTY;
  if (!ex->line_start->length())
    sql_ex.empty_flags |= LINE_START_EMPTY;
  if (!ex->escaped->length())
    sql_ex.empty_flags |= ESCAPED_EMPTY;
    
  skip_lines = ex->skip_lines;

  List_iterator<Item> li(fields_arg);
  field_lens_buf.length(0);
  fields_buf.length(0);
  Item* item;
  while ((item = li++))
  {
    num_fields++;
    uchar len= (uchar) item->name.length;
    field_block_len += len + 1;
    fields_buf.append(item->name.str, len + 1);
    field_lens_buf.append((char*)&len, 1);
  }

  field_lens = (const uchar*)field_lens_buf.ptr();
  fields = fields_buf.ptr();
}


/**
  Load_log_event::set_fields()

  @note
    This function can not use the member variable 
    for the database, since LOAD DATA INFILE on the slave
    can be for a different database than the current one.
    This is the reason for the affected_db argument to this method.
*/

void Load_log_event::set_fields(const char* affected_db, 
				List<Item> &field_list,
                                Name_resolution_context *context)
{
  uint i;
  const char* field = fields;
  for (i= 0; i < num_fields; i++)
  {
    LEX_CSTRING field_name= {field, field_lens[i] };
    field_list.push_back(new (thd->mem_root)
2769 2770 2771 2772
                         Item_field(thd, context,
                                    Lex_cstring_strlen(affected_db),
                                    Lex_cstring_strlen(table_name),
                                    field_name),
2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023
                         thd->mem_root);
    field+= field_lens[i]  + 1;
  }
}


#if defined(HAVE_REPLICATION)
/**
  Does the data loading job when executing a LOAD DATA on the slave.

  @param net
  @param rli
  @param use_rli_only_for_errors     If set to 1, rli is provided to
                                     Load_log_event::exec_event only for this
                                     function to have RPL_LOG_NAME and
                                     rli->last_slave_error, both being used by
                                     error reports. rli's position advancing
                                     is skipped (done by the caller which is
                                     Execute_load_log_event::exec_event).
                                     If set to 0, rli is provided for full use,
                                     i.e. for error reports and position
                                     advancing.

  @todo
    fix this; this can be done by testing rules in
    Create_file_log_event::exec_event() and then discarding Append_block and
    al.
  @todo
    this is a bug - this needs to be moved to the I/O thread

  @retval
    0           Success
  @retval
    1           Failure
*/

int Load_log_event::do_apply_event(NET* net, rpl_group_info *rgi,
                                   bool use_rli_only_for_errors)
{
  Relay_log_info const *rli= rgi->rli;
  Rpl_filter *rpl_filter= rli->mi->rpl_filter;
  DBUG_ENTER("Load_log_event::do_apply_event");

  DBUG_ASSERT(thd->query() == 0);
  set_thd_db(thd, rpl_filter, db, db_len);
  thd->clear_error(1);

  /* see Query_log_event::do_apply_event() and BUG#13360 */
  DBUG_ASSERT(!rgi->m_table_map.count());
  /*
    Usually lex_start() is called by mysql_parse(), but we need it here
    as the present method does not call mysql_parse().
  */
  lex_start(thd);
  thd->lex->local_file= local_fname;
  thd->reset_for_next_command(0);               // Errors are cleared above

   /*
    We test replicate_*_db rules. Note that we have already prepared
    the file to load, even if we are going to ignore and delete it
    now. So it is possible that we did a lot of disk writes for
    nothing. In other words, a big LOAD DATA INFILE on the master will
    still consume a lot of space on the slave (space in the relay log
    + space of temp files: twice the space of the file to load...)
    even if it will finally be ignored.  TODO: fix this; this can be
    done by testing rules in Create_file_log_event::do_apply_event()
    and then discarding Append_block and al. Another way is do the
    filtering in the I/O thread (more efficient: no disk writes at
    all).


    Note:   We do not need to execute reset_one_shot_variables() if this
            db_ok() test fails.
    Reason: The db stored in binlog events is the same for SET and for
            its companion query.  If the SET is ignored because of
            db_ok(), the companion query will also be ignored, and if
            the companion query is ignored in the db_ok() test of
            ::do_apply_event(), then the companion SET also have so
            we don't need to reset_one_shot_variables().
  */
  if (rpl_filter->db_ok(thd->db.str))
  {
    thd->set_time(when, when_sec_part);
    thd->set_query_id(next_query_id());
    thd->get_stmt_da()->opt_clear_warning_info(thd->query_id);

    TABLE_LIST tables;
    LEX_CSTRING db_name= { thd->strmake(thd->db.str, thd->db.length), thd->db.length };
    if (lower_case_table_names)
      my_casedn_str(system_charset_info, (char *)table_name);
    LEX_CSTRING tbl_name=   { table_name, strlen(table_name) };
    tables.init_one_table(&db_name, &tbl_name, 0, TL_WRITE);
    tables.updating= 1;

    // the table will be opened in mysql_load    
    if (rpl_filter->is_on() && !rpl_filter->tables_ok(thd->db.str, &tables))
    {
      // TODO: this is a bug - this needs to be moved to the I/O thread
      if (net)
        skip_load_data_infile(net);
    }
    else
    {
      enum enum_duplicates handle_dup;
      bool ignore= 0;
      char query_buffer[1024];
      String query_str(query_buffer, sizeof(query_buffer), system_charset_info);
      char *load_data_query;

      query_str.length(0);
      /*
        Forge LOAD DATA INFILE query which will be used in SHOW PROCESS LIST
        and written to slave's binlog if binlogging is on.
      */
      print_query(thd, FALSE, NULL, &query_str, NULL, NULL, NULL);
      if (!(load_data_query= (char *)thd->strmake(query_str.ptr(),
                                                  query_str.length())))
      {
        /*
          This will set thd->fatal_error in case of OOM. So we surely will notice
          that something is wrong.
        */
        goto error;
      }

      thd->set_query(load_data_query, (uint) (query_str.length()));

      if (sql_ex.opt_flags & REPLACE_FLAG)
        handle_dup= DUP_REPLACE;
      else if (sql_ex.opt_flags & IGNORE_FLAG)
      {
        ignore= 1;
        handle_dup= DUP_ERROR;
      }
      else
      {
        /*
          When replication is running fine, if it was DUP_ERROR on the
          master then we could choose IGNORE here, because if DUP_ERROR
          suceeded on master, and data is identical on the master and slave,
          then there should be no uniqueness errors on slave, so IGNORE is
          the same as DUP_ERROR. But in the unlikely case of uniqueness errors
          (because the data on the master and slave happen to be different
          (user error or bug), we want LOAD DATA to print an error message on
          the slave to discover the problem.

          If reading from net (a 3.23 master), mysql_load() will change this
          to IGNORE.
        */
        handle_dup= DUP_ERROR;
      }
      /*
        We need to set thd->lex->sql_command and thd->lex->duplicates
        since InnoDB tests these variables to decide if this is a LOAD
        DATA ... REPLACE INTO ... statement even though mysql_parse()
        is not called.  This is not needed in 5.0 since there the LOAD
        DATA ... statement is replicated using mysql_parse(), which
        sets the thd->lex fields correctly.
      */
      thd->lex->sql_command= SQLCOM_LOAD;
      thd->lex->duplicates= handle_dup;

      sql_exchange ex((char*)fname, sql_ex.opt_flags & DUMPFILE_FLAG);
      String field_term(sql_ex.field_term,sql_ex.field_term_len,log_cs);
      String enclosed(sql_ex.enclosed,sql_ex.enclosed_len,log_cs);
      String line_term(sql_ex.line_term,sql_ex.line_term_len,log_cs);
      String line_start(sql_ex.line_start,sql_ex.line_start_len,log_cs);
      String escaped(sql_ex.escaped,sql_ex.escaped_len, log_cs);
      ex.field_term= &field_term;
      ex.enclosed= &enclosed;
      ex.line_term= &line_term;
      ex.line_start= &line_start;
      ex.escaped= &escaped;

      ex.opt_enclosed = (sql_ex.opt_flags & OPT_ENCLOSED_FLAG);
      if (sql_ex.empty_flags & FIELD_TERM_EMPTY)
        ex.field_term->length(0);

      ex.skip_lines = skip_lines;
      List<Item> field_list;
      thd->lex->first_select_lex()->context.resolve_in_table_list_only(&tables);
      set_fields(tables.db.str,
                 field_list, &thd->lex->first_select_lex()->context);
      thd->variables.pseudo_thread_id= thread_id;
      if (net)
      {
        // mysql_load will use thd->net to read the file
        thd->net.vio = net->vio;
        // Make sure the client does not get confused about the packet sequence
        thd->net.pkt_nr = net->pkt_nr;
      }
      /*
        It is safe to use tmp_list twice because we are not going to
        update it inside mysql_load().
      */
      List<Item> tmp_list;
      if (thd->open_temporary_tables(&tables) ||
          mysql_load(thd, &ex, &tables, field_list, tmp_list, tmp_list,
                     handle_dup, ignore, net != 0))
        thd->is_slave_error= 1;
      if (thd->cuted_fields)
      {
        /* log_pos is the position of the LOAD event in the master log */
        sql_print_warning("Slave: load data infile on table '%s' at "
                          "log position %llu in log '%s' produced %ld "
                          "warning(s). Default database: '%s'",
                          (char*) table_name, log_pos, RPL_LOG_NAME,
                          (ulong) thd->cuted_fields,
                          thd->get_db());
      }
      if (net)
        net->pkt_nr= thd->net.pkt_nr;
    }
  }
  else
  {
    /*
      We will just ask the master to send us /dev/null if we do not
      want to load the data.
      TODO: this a bug - needs to be done in I/O thread
    */
    if (net)
      skip_load_data_infile(net);
  }

error:
  thd->net.vio = 0; 
  const char *remember_db= thd->get_db();
  thd->catalog= 0;
  thd->set_db(&null_clex_str);     /* will free the current database */
  thd->reset_query();
  thd->get_stmt_da()->set_overwrite_status(true);
  thd->is_error() ? trans_rollback_stmt(thd) : trans_commit_stmt(thd);
  thd->variables.option_bits&= ~(OPTION_BEGIN | OPTION_GTID_BEGIN);
  thd->get_stmt_da()->set_overwrite_status(false);
  close_thread_tables(thd);
  /*
    - If transaction rollback was requested due to deadlock
      perform it and release metadata locks.
    - If inside a multi-statement transaction,
    defer the release of metadata locks until the current
    transaction is either committed or rolled back. This prevents
    other statements from modifying the table for the entire
    duration of this transaction.  This provides commit ordering
    and guarantees serializability across multiple transactions.
    - If in autocommit mode, or outside a transactional context,
    automatically release metadata locks of the current statement.
  */
  if (thd->transaction_rollback_request)
  {
    trans_rollback_implicit(thd);
Marko Mäkelä's avatar
Marko Mäkelä committed
3024
    thd->release_transactional_locks();
3025 3026
  }
  else if (! thd->in_multi_stmt_transaction_mode())
Marko Mäkelä's avatar
Marko Mäkelä committed
3027
    thd->release_transactional_locks();
3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104
  else
    thd->mdl_context.release_statement_locks();

  DBUG_EXECUTE_IF("LOAD_DATA_INFILE_has_fatal_error",
                  thd->is_slave_error= 0; thd->is_fatal_error= 1;);

  if (unlikely(thd->is_slave_error))
  {
    /* this err/sql_errno code is copy-paste from net_send_error() */
    const char *err;
    int sql_errno;
    if (thd->is_error())
    {
      err= thd->get_stmt_da()->message();
      sql_errno= thd->get_stmt_da()->sql_errno();
    }
    else
    {
      sql_errno=ER_UNKNOWN_ERROR;
      err= ER_THD(thd, sql_errno);
    }
    rli->report(ERROR_LEVEL, sql_errno, rgi->gtid_info(), "\
Error '%s' running LOAD DATA INFILE on table '%s'. Default database: '%s'",
                    err, (char*)table_name, remember_db);
    free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));
    DBUG_RETURN(1);
  }
  free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));

  if (unlikely(thd->is_fatal_error))
  {
    char buf[256];
    my_snprintf(buf, sizeof(buf),
                "Running LOAD DATA INFILE on table '%-.64s'."
                " Default database: '%-.64s'",
                (char*)table_name,
                remember_db);

    rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR, rgi->gtid_info(),
                ER_THD(thd, ER_SLAVE_FATAL_ERROR), buf);
    DBUG_RETURN(1);
  }

  DBUG_RETURN( use_rli_only_for_errors ? 0 : Log_event::do_apply_event(rgi) );
}
#endif


/**************************************************************************
  Rotate_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void Rotate_log_event::pack_info(Protocol *protocol)
{
  StringBuffer<256> tmp(log_cs);
  tmp.length(0);
  tmp.append(new_log_ident, ident_len);
  tmp.append(STRING_WITH_LEN(";pos="));
  tmp.append_ulonglong(pos);
  protocol->store(tmp.ptr(), tmp.length(), &my_charset_bin);
}
#endif


Rotate_log_event::Rotate_log_event(const char* new_log_ident_arg,
                                   uint ident_len_arg, ulonglong pos_arg,
                                   uint flags_arg)
  :Log_event(), new_log_ident(new_log_ident_arg),
   pos(pos_arg),ident_len(ident_len_arg ? ident_len_arg :
                          (uint) strlen(new_log_ident_arg)), flags(flags_arg)
{
  DBUG_ENTER("Rotate_log_event::Rotate_log_event(...,flags)");
  DBUG_PRINT("enter",("new_log_ident: %s  pos: %llu  flags: %lu", new_log_ident_arg,
                      pos_arg, (ulong) flags));
  cache_type= EVENT_NO_CACHE;
  if (flags & DUP_NAME)
3105
    new_log_ident= my_strndup(PSI_INSTRUMENT_ME, new_log_ident_arg, ident_len, MYF(MY_WME));
3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260
  if (flags & RELAY_LOG)
    set_relay_log_event();
  DBUG_VOID_RETURN;
}


bool Rotate_log_event::write()
{
  char buf[ROTATE_HEADER_LEN];
  int8store(buf + R_POS_OFFSET, pos);
  return (write_header(ROTATE_HEADER_LEN + ident_len) ||
          write_data(buf, ROTATE_HEADER_LEN) ||
          write_data(new_log_ident, (uint) ident_len) ||
          write_footer());
}


#if defined(HAVE_REPLICATION)

/*
  Got a rotate log event from the master.

  This is mainly used so that we can later figure out the logname and
  position for the master.

  We can't rotate the slave's BINlog as this will cause infinitive rotations
  in a A -> B -> A setup.
  The NOTES below is a wrong comment which will disappear when 4.1 is merged.

  This must only be called from the Slave SQL thread, since it calls
  Relay_log_info::flush().

  @retval
    0	ok
    1   error
*/
int Rotate_log_event::do_update_pos(rpl_group_info *rgi)
{
  int error= 0;
  Relay_log_info *rli= rgi->rli;
  DBUG_ENTER("Rotate_log_event::do_update_pos");

  DBUG_PRINT("info", ("server_id=%lu; ::server_id=%lu",
                      (ulong) this->server_id, (ulong) global_system_variables.server_id));
  DBUG_PRINT("info", ("new_log_ident: %s", this->new_log_ident));
  DBUG_PRINT("info", ("pos: %llu", this->pos));

  /*
    If we are in a transaction or in a group: the only normal case is
    when the I/O thread was copying a big transaction, then it was
    stopped and restarted: we have this in the relay log:

    BEGIN
    ...
    ROTATE (a fake one)
    ...
    COMMIT or ROLLBACK

    In that case, we don't want to touch the coordinates which
    correspond to the beginning of the transaction.  Starting from
    5.0.0, there also are some rotates from the slave itself, in the
    relay log, which shall not change the group positions.

    In parallel replication, rotate event is executed out-of-band with normal
    events, so we cannot update group_master_log_name or _pos here, it will
    be updated with the next normal event instead.
  */
  if ((server_id != global_system_variables.server_id ||
       rli->replicate_same_server_id) &&
      !is_relay_log_event() &&
      !rli->is_in_group() &&
      !rgi->is_parallel_exec)
  {
    mysql_mutex_lock(&rli->data_lock);
    DBUG_PRINT("info", ("old group_master_log_name: '%s'  "
                        "old group_master_log_pos: %lu",
                        rli->group_master_log_name,
                        (ulong) rli->group_master_log_pos));
    memcpy(rli->group_master_log_name, new_log_ident, ident_len+1);
    rli->notify_group_master_log_name_update();
    rli->inc_group_relay_log_pos(pos, rgi, TRUE /* skip_lock */);
    DBUG_PRINT("info", ("new group_master_log_name: '%s'  "
                        "new group_master_log_pos: %lu",
                        rli->group_master_log_name,
                        (ulong) rli->group_master_log_pos));
    mysql_mutex_unlock(&rli->data_lock);
    rpl_global_gtid_slave_state->record_and_update_gtid(thd, rgi);
    error= rli->flush();
    
    /*
      Reset thd->variables.option_bits and sql_mode etc, because this could
      be the signal of a master's downgrade from 5.0 to 4.0.
      However, no need to reset description_event_for_exec: indeed, if the next
      master is 5.0 (even 5.0.1) we will soon get a Format_desc; if the next
      master is 4.0 then the events are in the slave's format (conversion).
    */
    set_slave_thread_options(thd);
    set_slave_thread_default_charset(thd, rgi);
    thd->variables.sql_mode= global_system_variables.sql_mode;
    thd->variables.auto_increment_increment=
      thd->variables.auto_increment_offset= 1;
  }
  else
    rgi->inc_event_relay_log_pos();

  DBUG_RETURN(error);
}


Log_event::enum_skip_reason
Rotate_log_event::do_shall_skip(rpl_group_info *rgi)
{
  enum_skip_reason reason= Log_event::do_shall_skip(rgi);

  switch (reason) {
  case Log_event::EVENT_SKIP_NOT:
  case Log_event::EVENT_SKIP_COUNT:
    return Log_event::EVENT_SKIP_NOT;

  case Log_event::EVENT_SKIP_IGNORE:
    return Log_event::EVENT_SKIP_IGNORE;
  }
  DBUG_ASSERT(0);
  return Log_event::EVENT_SKIP_NOT;             // To keep compiler happy
}

#endif


/**************************************************************************
  Binlog_checkpoint_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void Binlog_checkpoint_log_event::pack_info(Protocol *protocol)
{
  protocol->store(binlog_file_name, binlog_file_len, &my_charset_bin);
}


Log_event::enum_skip_reason
Binlog_checkpoint_log_event::do_shall_skip(rpl_group_info *rgi)
{
  enum_skip_reason reason= Log_event::do_shall_skip(rgi);
  if (reason == EVENT_SKIP_COUNT)
    reason= EVENT_SKIP_NOT;
  return reason;
}
#endif


Binlog_checkpoint_log_event::Binlog_checkpoint_log_event(
        const char *binlog_file_name_arg,
        uint binlog_file_len_arg)
  :Log_event(),
3261
   binlog_file_name(my_strndup(PSI_INSTRUMENT_ME, binlog_file_name_arg, binlog_file_len_arg,
3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286
                               MYF(MY_WME))),
   binlog_file_len(binlog_file_len_arg)
{
  cache_type= EVENT_NO_CACHE;
}


bool Binlog_checkpoint_log_event::write()
{
  uchar buf[BINLOG_CHECKPOINT_HEADER_LEN];
  int4store(buf, binlog_file_len);
  return write_header(BINLOG_CHECKPOINT_HEADER_LEN + binlog_file_len) ||
         write_data(buf, BINLOG_CHECKPOINT_HEADER_LEN) ||
         write_data(binlog_file_name, binlog_file_len) ||
         write_footer();
}


/**************************************************************************
        Global transaction ID stuff
**************************************************************************/

Gtid_log_event::Gtid_log_event(THD *thd_arg, uint64 seq_no_arg,
                               uint32 domain_id_arg, bool standalone,
                               uint16 flags_arg, bool is_transactional,
3287 3288
                               uint64 commit_id_arg, bool has_xid,
                               bool ro_1pc)
3289 3290
  : Log_event(thd_arg, flags_arg, is_transactional),
    seq_no(seq_no_arg), commit_id(commit_id_arg), domain_id(domain_id_arg),
3291 3292 3293
    flags2((standalone ? FL_STANDALONE : 0) |
           (commit_id_arg ? FL_GROUP_COMMIT_ID : 0)),
    flags_extra(0), extra_engines(0)
3294 3295
{
  cache_type= Log_event::EVENT_NO_CACHE;
Marko Mäkelä's avatar
Marko Mäkelä committed
3296
  bool is_tmp_table= thd_arg->lex->stmt_accessed_temp_table();
3297 3298
  if (thd_arg->transaction->stmt.trans_did_wait() ||
      thd_arg->transaction->all.trans_did_wait())
3299
    flags2|= FL_WAITED;
3300 3301
  if (thd_arg->transaction->stmt.trans_did_ddl() ||
      thd_arg->transaction->stmt.has_created_dropped_temp_table() ||
Marko Mäkelä's avatar
Marko Mäkelä committed
3302
      thd_arg->transaction->stmt.trans_executed_admin_cmd() ||
3303
      thd_arg->transaction->all.trans_did_ddl() ||
Marko Mäkelä's avatar
Marko Mäkelä committed
3304 3305
      thd_arg->transaction->all.has_created_dropped_temp_table() ||
      thd_arg->transaction->all.trans_executed_admin_cmd())
3306
    flags2|= FL_DDL;
3307 3308 3309 3310
  else if (is_transactional && !is_tmp_table &&
           !(thd_arg->transaction->all.modified_non_trans_table &&
             thd->variables.binlog_direct_non_trans_update == 0 &&
             !thd->is_current_stmt_binlog_format_row()))
3311 3312 3313 3314 3315 3316
    flags2|= FL_TRANSACTIONAL;
  if (!(thd_arg->variables.option_bits & OPTION_RPL_SKIP_PARALLEL))
    flags2|= FL_ALLOW_PARALLEL;
  /* Preserve any DDL or WAITED flag in the slave's binlog. */
  if (thd_arg->rgi_slave)
    flags2|= (thd_arg->rgi_slave->gtid_ev_flags2 & (FL_DDL|FL_WAITED));
3317

3318
  XID_STATE &xid_state= thd->transaction->xid_state;
3319
  if (is_transactional)
3320
  {
3321 3322 3323 3324
    if (xid_state.is_explicit_XA() &&
        (thd->lex->sql_command == SQLCOM_XA_PREPARE ||
         xid_state.get_state_code() == XA_PREPARED))
    {
Marko Mäkelä's avatar
Marko Mäkelä committed
3325 3326
      DBUG_ASSERT(!(thd->lex->sql_command == SQLCOM_XA_COMMIT &&
                    thd->lex->xa_opt == XA_ONE_PHASE));
3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347

      flags2|= thd->lex->sql_command == SQLCOM_XA_PREPARE ?
        FL_PREPARED_XA : FL_COMPLETED_XA;
      xid.set(xid_state.get_xid());
    }
    /* count non-zero extra recoverable engines; total = extra + 1 */
    if (has_xid)
    {
      DBUG_ASSERT(ha_count_rw_2pc(thd_arg,
                                  thd_arg->in_multi_stmt_transaction_mode()));

      extra_engines=
        ha_count_rw_2pc(thd_arg, thd_arg->in_multi_stmt_transaction_mode()) - 1;
    }
    else if (ro_1pc)
    {
      extra_engines= UCHAR_MAX;
    }
    else if (thd->lex->sql_command == SQLCOM_XA_PREPARE)
    {
      DBUG_ASSERT(thd_arg->in_multi_stmt_transaction_mode());
3348

3349 3350 3351 3352 3353
      uint8 count= ha_count_rw_2pc(thd_arg, true);
      extra_engines= count > 1 ? 0 : UCHAR_MAX;
    }
    if (extra_engines > 0)
      flags_extra|= FL_EXTRA_MULTI_ENGINE;
3354
  }
3355 3356 3357

  DBUG_ASSERT(thd_arg->lex->sql_command != SQLCOM_CREATE_SEQUENCE ||
              (flags2 & FL_DDL) || thd_arg->in_multi_stmt_transaction_mode());
3358 3359 3360 3361 3362 3363 3364 3365
}


/*
  Used to record GTID while sending binlog to slave, without having to
  fully contruct every Gtid_log_event() needlessly.
*/
bool
3366
Gtid_log_event::peek(const uchar *event_start, size_t event_len,
3367 3368 3369 3370
                     enum enum_binlog_checksum_alg checksum_alg,
                     uint32 *domain_id, uint32 *server_id, uint64 *seq_no,
                     uchar *flags2, const Format_description_log_event *fdev)
{
3371
  const uchar *p;
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391

  if (checksum_alg == BINLOG_CHECKSUM_ALG_CRC32)
  {
    if (event_len > BINLOG_CHECKSUM_LEN)
      event_len-= BINLOG_CHECKSUM_LEN;
    else
      event_len= 0;
  }
  else
    DBUG_ASSERT(checksum_alg == BINLOG_CHECKSUM_ALG_UNDEF ||
                checksum_alg == BINLOG_CHECKSUM_ALG_OFF);

  if (event_len < (uint32)fdev->common_header_len + GTID_HEADER_LEN)
    return true;
  *server_id= uint4korr(event_start + SERVER_ID_OFFSET);
  p= event_start + fdev->common_header_len;
  *seq_no= uint8korr(p);
  p+= 8;
  *domain_id= uint4korr(p);
  p+= 4;
3392
  *flags2= *p;
3393 3394 3395 3396 3397 3398 3399
  return false;
}


bool
Gtid_log_event::write()
{
3400 3401
  uchar buf[GTID_HEADER_LEN+2+sizeof(XID) + /* flags_extra: */ 1+4];
  size_t write_len= 13;
3402 3403 3404 3405 3406 3407

  int8store(buf, seq_no);
  int4store(buf+8, domain_id);
  buf[12]= flags2;
  if (flags2 & FL_GROUP_COMMIT_ID)
  {
3408 3409 3410
    DBUG_ASSERT(write_len + 8 == GTID_HEADER_LEN + 2);

    int8store(buf+write_len, commit_id);
3411 3412
    write_len= GTID_HEADER_LEN + 2;
  }
3413

3414 3415 3416 3417 3418
  if (flags2 & (FL_PREPARED_XA | FL_COMPLETED_XA)
#ifndef DBUG_OFF
      && DBUG_EVALUATE_IF("negate_xid_from_gtid", 0, 1)
#endif
  )
3419
  {
3420 3421 3422 3423 3424
    int4store(&buf[write_len],   xid.formatID);
    buf[write_len +4]=   (uchar) xid.gtrid_length;
    buf[write_len +4+1]= (uchar) xid.bqual_length;
    write_len+= 6;
    long data_length= xid.bqual_length + xid.gtrid_length;
3425 3426 3427 3428 3429

#ifndef DBUG_OFF
    if (DBUG_EVALUATE_IF("negate_xid_data_from_gtid", 0, 1))
    {
#endif
3430 3431
    memcpy(buf+write_len, xid.data, data_length);
    write_len+= data_length;
3432 3433 3434
#ifndef DBUG_OFF
    }
#endif
3435
  }
3436 3437 3438 3439

  DBUG_EXECUTE_IF("inject_fl_extra_multi_engine_into_gtid", {
    flags_extra|= FL_EXTRA_MULTI_ENGINE;
  });
3440 3441 3442 3443 3444
  if (flags_extra > 0)
  {
    buf[write_len]= flags_extra;
    write_len++;
  }
3445 3446 3447 3448
  DBUG_EXECUTE_IF("inject_fl_extra_multi_engine_into_gtid", {
    flags_extra&= ~FL_EXTRA_MULTI_ENGINE;
  });

3449 3450 3451 3452 3453
  if (flags_extra & FL_EXTRA_MULTI_ENGINE)
  {
    buf[write_len]= extra_engines;
    write_len++;
  }
3454 3455 3456 3457

  if (write_len < GTID_HEADER_LEN)
  {
    bzero(buf+write_len, GTID_HEADER_LEN-write_len);
3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
    write_len= GTID_HEADER_LEN;
  }
  return write_header(write_len) ||
         write_data(buf, write_len) ||
         write_footer();
}


/*
  Replace a GTID event with either a BEGIN event, dummy event, or nothing, as
  appropriate to work with old slave that does not know global transaction id.

  The need_dummy_event argument is an IN/OUT argument. It is passed as TRUE
  if slave has capability lower than MARIA_SLAVE_CAPABILITY_TOLERATE_HOLES.
  It is returned TRUE if we return a BEGIN (or dummy) event to be sent to the
  slave, FALSE if event should be skipped completely.
*/
int
Gtid_log_event::make_compatible_event(String *packet, bool *need_dummy_event,
                                      ulong ev_offset,
                                      enum enum_binlog_checksum_alg checksum_alg)
{
  uchar flags2;
  if (packet->length() - ev_offset < LOG_EVENT_HEADER_LEN + GTID_HEADER_LEN)
    return 1;
  flags2= (*packet)[ev_offset + LOG_EVENT_HEADER_LEN + 12];
  if (flags2 & FL_STANDALONE)
  {
    if (*need_dummy_event)
      return Query_log_event::dummy_event(packet, ev_offset, checksum_alg);
    return 0;
  }

  *need_dummy_event= true;
  return Query_log_event::begin_event(packet, ev_offset, checksum_alg);
}


#ifdef HAVE_REPLICATION
void
Gtid_log_event::pack_info(Protocol *protocol)
{
3500
  char buf[6+5+10+1+10+1+20+1+4+20+1+ ser_buf_size+5 /* sprintf */];
3501
  char *p;
3502 3503 3504 3505 3506 3507
  p = strmov(buf, (flags2 & FL_STANDALONE  ? "GTID " :
                   flags2 & FL_PREPARED_XA ? "XA START " : "BEGIN GTID "));
  if (flags2 & FL_PREPARED_XA)
  {
    p+= sprintf(p, "%s GTID ", xid.serialize());
  }
3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567
  p= longlong10_to_str(domain_id, p, 10);
  *p++= '-';
  p= longlong10_to_str(server_id, p, 10);
  *p++= '-';
  p= longlong10_to_str(seq_no, p, 10);
  if (flags2 & FL_GROUP_COMMIT_ID)
  {
    p= strmov(p, " cid=");
    p= longlong10_to_str(commit_id, p, 10);
  }

  protocol->store(buf, p-buf, &my_charset_bin);
}

static char gtid_begin_string[] = "BEGIN";

int
Gtid_log_event::do_apply_event(rpl_group_info *rgi)
{
  ulonglong bits= thd->variables.option_bits;
  thd->variables.server_id= this->server_id;
  thd->variables.gtid_domain_id= this->domain_id;
  thd->variables.gtid_seq_no= this->seq_no;
  rgi->gtid_ev_flags2= flags2;
  thd->reset_for_next_command();

  if (opt_gtid_strict_mode && opt_bin_log && opt_log_slave_updates)
  {
    if (mysql_bin_log.check_strict_gtid_sequence(this->domain_id,
                                                 this->server_id, this->seq_no))
      return 1;
  }

  DBUG_ASSERT((bits & OPTION_GTID_BEGIN) == 0);

  Master_info *mi=rgi->rli->mi;
  switch (flags2 & (FL_DDL | FL_TRANSACTIONAL))
  {
    case FL_TRANSACTIONAL:
      mi->total_trans_groups++;
      break;
    case FL_DDL:
      mi->total_ddl_groups++;
    break;
    default:
      mi->total_non_trans_groups++;
  }

  if (flags2 & FL_STANDALONE)
    return 0;

  /* Execute this like a BEGIN query event. */
  bits|= OPTION_GTID_BEGIN;
  if (flags2 & FL_ALLOW_PARALLEL)
    bits&= ~(ulonglong)OPTION_RPL_SKIP_PARALLEL;
  else
    bits|= (ulonglong)OPTION_RPL_SKIP_PARALLEL;
  thd->variables.option_bits= bits;
  DBUG_PRINT("info", ("Set OPTION_GTID_BEGIN"));
  thd->is_slave_error= 0;
3568 3569 3570

  char buf_xa[sizeof("XA START") + 1 + ser_buf_size];
  if (flags2 & FL_PREPARED_XA)
3571
  {
3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584
    const char fmt[]= "XA START %s";

    thd->lex->xid= &xid;
    thd->lex->xa_opt= XA_NONE;
    sprintf(buf_xa, fmt, xid.serialize());
    thd->set_query_and_id(buf_xa, static_cast<uint32>(strlen(buf_xa)),
                          &my_charset_bin, next_query_id());
    thd->lex->sql_command= SQLCOM_XA_START;
    if (trans_xa_start(thd))
    {
      DBUG_PRINT("error", ("trans_xa_start() failed"));
      thd->is_slave_error= 1;
    }
3585
  }
3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
  else
  {
    thd->set_query_and_id(gtid_begin_string, sizeof(gtid_begin_string)-1,
                          &my_charset_bin, next_query_id());
    thd->lex->sql_command= SQLCOM_BEGIN;
    if (trans_begin(thd, 0))
    {
      DBUG_PRINT("error", ("trans_begin() failed"));
      thd->is_slave_error= 1;
    }
  }
  status_var_increment(thd->status_var.com_stat[thd->lex->sql_command]);
3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652
  thd->update_stats();

  if (likely(!thd->is_slave_error))
    general_log_write(thd, COM_QUERY, thd->query(), thd->query_length());

  thd->reset_query();
  free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));
  return thd->is_slave_error;
}


int
Gtid_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}


Log_event::enum_skip_reason
Gtid_log_event::do_shall_skip(rpl_group_info *rgi)
{
  Relay_log_info *rli= rgi->rli;
  /*
    An event skipped due to @@skip_replication must not be counted towards the
    number of events to be skipped due to @@sql_slave_skip_counter.
  */
  if (flags & LOG_EVENT_SKIP_REPLICATION_F &&
      opt_replicate_events_marked_for_skip != RPL_SKIP_REPLICATE)
    return Log_event::EVENT_SKIP_IGNORE;

  if (rli->slave_skip_counter > 0)
  {
    if (!(flags2 & FL_STANDALONE))
    {
      thd->variables.option_bits|= OPTION_BEGIN;
      DBUG_ASSERT(rgi->rli->get_flag(Relay_log_info::IN_TRANSACTION));
    }
    return Log_event::continue_group(rgi);
  }
  return Log_event::do_shall_skip(rgi);
}


#endif  /* HAVE_REPLICATION */



Gtid_list_log_event::Gtid_list_log_event(rpl_binlog_state *gtid_set,
                                         uint32 gl_flags_)
  : count(gtid_set->count()), gl_flags(gl_flags_), list(0), sub_id_list(0)
{
  cache_type= EVENT_NO_CACHE;
  /* Failure to allocate memory will be caught by is_valid() returning false. */
  if (count < (1<<28) &&
3653 3654
      (list = (rpl_gtid *)my_malloc(PSI_INSTRUMENT_ME,
                           count * sizeof(*list) + (count == 0), MYF(MY_WME))))
3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665
    gtid_set->get_gtid_list(list, count);
}


Gtid_list_log_event::Gtid_list_log_event(slave_connection_state *gtid_set,
                                         uint32 gl_flags_)
  : count(gtid_set->count()), gl_flags(gl_flags_), list(0), sub_id_list(0)
{
  cache_type= EVENT_NO_CACHE;
  /* Failure to allocate memory will be caught by is_valid() returning false. */
  if (count < (1<<28) &&
3666 3667
      (list = (rpl_gtid *)my_malloc(PSI_INSTRUMENT_ME,
                          count * sizeof(*list) + (count == 0), MYF(MY_WME))))
3668 3669 3670 3671 3672 3673 3674
  {
    gtid_set->get_gtid_list(list, count);
#if defined(HAVE_REPLICATION)
    if (gl_flags & FLAG_IGN_GTIDS)
    {
      uint32 i;

3675 3676
      if (!(sub_id_list= (uint64 *)my_malloc(PSI_INSTRUMENT_ME,
                                         count * sizeof(uint64), MYF(MY_WME))))
3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980
      {
        my_free(list);
        list= NULL;
        return;
      }
      for (i= 0; i < count; ++i)
      {
        if (!(sub_id_list[i]=
              rpl_global_gtid_slave_state->next_sub_id(list[i].domain_id)))
        {
          my_free(list);
          my_free(sub_id_list);
          list= NULL;
          sub_id_list= NULL;
          return;
        }
      }
    }
#endif
  }
}


#if defined(HAVE_REPLICATION)
bool
Gtid_list_log_event::to_packet(String *packet)
{
  uint32 i;
  uchar *p;
  uint32 needed_length;

  DBUG_ASSERT(count < 1<<28);

  needed_length= packet->length() + get_data_size();
  if (packet->reserve(needed_length))
    return true;
  p= (uchar *)packet->ptr() + packet->length();;
  packet->length(needed_length);
  int4store(p, (count & ((1<<28)-1)) | gl_flags);
  p += 4;
  /* Initialise the padding for empty Gtid_list. */
  if (count == 0)
    int2store(p, 0);
  for (i= 0; i < count; ++i)
  {
    int4store(p, list[i].domain_id);
    int4store(p+4, list[i].server_id);
    int8store(p+8, list[i].seq_no);
    p += 16;
  }

  return false;
}


bool
Gtid_list_log_event::write()
{
  char buf[128];
  String packet(buf, sizeof(buf), system_charset_info);

  packet.length(0);
  if (to_packet(&packet))
    return true;
  return write_header(get_data_size()) ||
         write_data(packet.ptr(), packet.length()) ||
         write_footer();
}


int
Gtid_list_log_event::do_apply_event(rpl_group_info *rgi)
{
  Relay_log_info *rli= const_cast<Relay_log_info*>(rgi->rli);
  int ret;
  if (gl_flags & FLAG_IGN_GTIDS)
  {
    void *hton= NULL;
    uint32 i;

    for (i= 0; i < count; ++i)
    {
      if ((ret= rpl_global_gtid_slave_state->record_gtid(thd, &list[i],
                                                         sub_id_list[i],
                                                         false, false, &hton)))
        return ret;
      rpl_global_gtid_slave_state->update_state_hash(sub_id_list[i], &list[i],
                                                     hton, NULL);
    }
  }
  ret= Log_event::do_apply_event(rgi);
  if (rli->until_condition == Relay_log_info::UNTIL_GTID &&
      (gl_flags & FLAG_UNTIL_REACHED))
  {
    char str_buf[128];
    String str(str_buf, sizeof(str_buf), system_charset_info);
    rli->until_gtid_pos.to_string(&str);
    sql_print_information("Slave SQL thread stops because it reached its"
                          " UNTIL master_gtid_pos %s", str.c_ptr_safe());
    rli->abort_slave= true;
    rli->stop_for_until= true;
  }
  free_root(thd->mem_root, MYF(MY_KEEP_PREALLOC));
  return ret;
}


Log_event::enum_skip_reason
Gtid_list_log_event::do_shall_skip(rpl_group_info *rgi)
{
  enum_skip_reason reason= Log_event::do_shall_skip(rgi);
  if (reason == EVENT_SKIP_COUNT)
    reason= EVENT_SKIP_NOT;
  return reason;
}


void
Gtid_list_log_event::pack_info(Protocol *protocol)
{
  char buf_mem[1024];
  String buf(buf_mem, sizeof(buf_mem), system_charset_info);
  uint32 i;
  bool first;

  buf.length(0);
  buf.append(STRING_WITH_LEN("["));
  first= true;
  for (i= 0; i < count; ++i)
    rpl_slave_state_tostring_helper(&buf, &list[i], &first);
  buf.append(STRING_WITH_LEN("]"));

  protocol->store(&buf);
}
#endif  /* HAVE_REPLICATION */



/**************************************************************************
	Intvar_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void Intvar_log_event::pack_info(Protocol *protocol)
{
  char buf[256], *pos;
  pos= strmake(buf, get_var_type_name(), sizeof(buf)-23);
  *pos++= '=';
  pos= longlong10_to_str(val, pos, -10);
  protocol->store(buf, (uint) (pos-buf), &my_charset_bin);
}
#endif


bool Intvar_log_event::write()
{
  uchar buf[9];
  buf[I_TYPE_OFFSET]= (uchar) type;
  int8store(buf + I_VAL_OFFSET, val);
  return write_header(sizeof(buf)) ||
         write_data(buf, sizeof(buf)) ||
         write_footer();
}


#if defined(HAVE_REPLICATION)

/*
  Intvar_log_event::do_apply_event()
*/

int Intvar_log_event::do_apply_event(rpl_group_info *rgi)
{
  DBUG_ENTER("Intvar_log_event::do_apply_event");
  if (rgi->deferred_events_collecting)
  {
    DBUG_PRINT("info",("deferring event"));
    DBUG_RETURN(rgi->deferred_events->add(this));
  }

  switch (type) {
  case LAST_INSERT_ID_EVENT:
    thd->first_successful_insert_id_in_prev_stmt= val;
    DBUG_PRINT("info",("last_insert_id_event: %ld", (long) val));
    break;
  case INSERT_ID_EVENT:
    thd->force_one_auto_inc_interval(val);
    break;
  }
  DBUG_RETURN(0);
}

int Intvar_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}


Log_event::enum_skip_reason
Intvar_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    It is a common error to set the slave skip counter to 1 instead of
    2 when recovering from an insert which used a auto increment,
    rand, or user var.  Therefore, if the slave skip counter is 1, we
    just say that this event should be skipped by ignoring it, meaning
    that we do not change the value of the slave skip counter since it
    will be decreased by the following insert event.
  */
  return continue_group(rgi);
}

#endif


/**************************************************************************
  Rand_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void Rand_log_event::pack_info(Protocol *protocol)
{
  char buf1[256], *pos;
  pos= strmov(buf1,"rand_seed1=");
  pos= int10_to_str((long) seed1, pos, 10);
  pos= strmov(pos, ",rand_seed2=");
  pos= int10_to_str((long) seed2, pos, 10);
  protocol->store(buf1, (uint) (pos-buf1), &my_charset_bin);
}
#endif


bool Rand_log_event::write()
{
  uchar buf[16];
  int8store(buf + RAND_SEED1_OFFSET, seed1);
  int8store(buf + RAND_SEED2_OFFSET, seed2);
  return write_header(sizeof(buf)) ||
         write_data(buf, sizeof(buf)) ||
         write_footer();
}


#if defined(HAVE_REPLICATION)
int Rand_log_event::do_apply_event(rpl_group_info *rgi)
{
  if (rgi->deferred_events_collecting)
    return rgi->deferred_events->add(this);

  thd->rand.seed1= (ulong) seed1;
  thd->rand.seed2= (ulong) seed2;
  return 0;
}

int Rand_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}


Log_event::enum_skip_reason
Rand_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    It is a common error to set the slave skip counter to 1 instead of
    2 when recovering from an insert which used a auto increment,
    rand, or user var.  Therefore, if the slave skip counter is 1, we
    just say that this event should be skipped by ignoring it, meaning
    that we do not change the value of the slave skip counter since it
    will be decreased by the following insert event.
  */
  return continue_group(rgi);
}

/**
   Exec deferred Int-, Rand- and User- var events prefixing
   a Query-log-event event.

   @param thd THD handle

   @return false on success, true if a failure in an event applying occurred.
*/
bool slave_execute_deferred_events(THD *thd)
{
  bool res= false;
  rpl_group_info *rgi= thd->rgi_slave;

  DBUG_ASSERT(rgi && (!rgi->deferred_events_collecting || rgi->deferred_events));

  if (!rgi->deferred_events_collecting || rgi->deferred_events->is_empty())
    return res;

  res= rgi->deferred_events->execute(rgi);
  rgi->deferred_events->rewind();

  return res;
}

#endif /* HAVE_REPLICATION */


/**************************************************************************
3981
  Xid_apply_log_event methods
3982 3983 3984
**************************************************************************/

#if defined(HAVE_REPLICATION)
3985 3986

int Xid_apply_log_event::do_record_gtid(THD *thd, rpl_group_info *rgi,
3987 3988
                                        bool in_trans, void **out_hton,
                                        bool force_err)
3989
{
3990 3991
  int err= 0;
  Relay_log_info const *rli= rgi->rli;
3992

3993 3994 3995 3996
  rgi->gtid_pending= false;
  err= rpl_global_gtid_slave_state->record_gtid(thd, &rgi->current_gtid,
                                                rgi->gtid_sub_id,
                                                in_trans, false, out_hton);
3997

3998 3999 4000 4001 4002
  if (unlikely(err))
  {
    int ec= thd->get_stmt_da()->sql_errno();
    /*
      Do not report an error if this is really a kill due to a deadlock.
4003 4004 4005 4006
      In this case, the transaction will be re-tried instead. Unless force_err
      is set, as in the case of XA PREPARE, as the GTID state is updated as a
      separate transaction, and if that fails, we should not retry but exit in
      error immediately.
4007
    */
4008 4009 4010 4011 4012 4013 4014 4015 4016
    if (!is_parallel_retry_error(rgi, ec) || force_err)
    {
      char buff[MAX_SLAVE_ERRMSG];
      buff[0]= 0;
      aggregate_da_errors(buff, sizeof(buff), thd->get_stmt_da());

      if (force_err)
        thd->clear_error();

4017 4018
      rli->report(ERROR_LEVEL, ER_CANNOT_UPDATE_GTID_STATE, rgi->gtid_info(),
                  "Error during XID COMMIT: failed to update GTID state in "
4019
                  "%s.%s: %d: %s the event's master log %s, end_log_pos %llu",
4020
                  "mysql", rpl_gtid_slave_state_table_name.str, ec,
4021 4022
                  buff, RPL_LOG_NAME, log_pos);
    }
4023 4024
    thd->is_slave_error= 1;
  }
4025

4026 4027
  return err;
}
4028

4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041
static bool wsrep_must_replay(THD *thd)
{
#ifdef WITH_WSREP
  mysql_mutex_lock(&thd->LOCK_thd_data);
  bool res= WSREP(thd) && thd->wsrep_trx().state() == wsrep::transaction::s_must_replay;
  mysql_mutex_unlock(&thd->LOCK_thd_data);
  return res;
#else
  return false;
#endif
}


4042
int Xid_apply_log_event::do_apply_event(rpl_group_info *rgi)
4043 4044 4045 4046 4047
{
  bool res;
  int err;
  uint64 sub_id= 0;
  void *hton= NULL;
4048
  rpl_gtid gtid;
4049 4050

  /*
4051 4052 4053
    An instance of this class such as XID_EVENT works like a COMMIT
    statement. It updates mysql.gtid_slave_pos with the GTID of the
    current transaction.
4054 4055
    Therefore, it acts much like a normal SQL statement, so we need to do
    THD::reset_for_next_command() as if starting a new statement.
4056 4057 4058

    XA_PREPARE_LOG_EVENT also updates the gtid table *but* the update gets
    committed as separate "autocommit" transaction.
4059 4060 4061 4062 4063 4064 4065 4066 4067 4068
  */
  thd->reset_for_next_command();
  /*
    Record any GTID in the same transaction, so slave state is transactionally
    consistent.
  */
#ifdef WITH_WSREP
  thd->wsrep_affected_rows= 0;
#endif

4069 4070 4071
#ifndef DBUG_OFF
  bool record_gtid_delayed_for_xa= false;
#endif
4072 4073 4074 4075
  if (rgi->gtid_pending)
  {
    sub_id= rgi->gtid_sub_id;
    gtid= rgi->current_gtid;
4076

4077
    if (!thd->transaction->xid_state.is_explicit_XA())
4078
    {
4079 4080 4081 4082 4083 4084 4085 4086 4087 4088
      if ((err= do_record_gtid(thd, rgi, true /* in_trans */, &hton)))
        return err;

      DBUG_EXECUTE_IF("gtid_fail_after_record_gtid",
                      {
                        my_error(ER_ERROR_DURING_COMMIT, MYF(0),
                                 HA_ERR_WRONG_COMMAND);
                        thd->is_slave_error= 1;
                        return 1;
                      });
4089
    }
4090 4091 4092 4093
#ifndef DBUG_OFF
    else
      record_gtid_delayed_for_xa= true;
#endif
4094 4095
  }

4096
  general_log_print(thd, COM_QUERY, get_query());
4097
  thd->variables.option_bits&= ~OPTION_GTID_BEGIN;
4098 4099 4100
  res= do_commit();
  if (!res && rgi->gtid_pending)
  {
4101
    DBUG_ASSERT(!thd->transaction->xid_state.is_explicit_XA());
4102

4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118
    DBUG_ASSERT(record_gtid_delayed_for_xa);
    if (thd->rgi_slave->is_parallel_exec)
    {
      /*
        With XA, since the transaction is prepared/committed without updating
        the GTID pos (MDEV-32020...), we need here to clear any pending
        deadlock kill.

        Otherwise if the kill happened after the prepare/commit completed, it
        might end up killing the subsequent GTID position update, causing the
        slave to fail with error.
      */
      wait_for_pending_deadlock_kill(thd, thd->rgi_slave);
      thd->reset_killed();
    }

4119
    if ((err= do_record_gtid(thd, rgi, false, &hton, true)))
4120 4121 4122
      return err;
  }

4123
  if (sub_id && (!res || wsrep_must_replay(thd)))
4124 4125 4126 4127
    rpl_global_gtid_slave_state->update_state_hash(sub_id, &gtid, hton, rgi);
  /*
    Increment the global status commit count variable
  */
4128 4129
  enum enum_sql_command cmd= !thd->transaction->xid_state.is_explicit_XA()
    ? SQLCOM_COMMIT : SQLCOM_XA_PREPARE;
4130
  status_var_increment(thd->status_var.com_stat[cmd]);
4131 4132 4133 4134 4135

  return res;
}

Log_event::enum_skip_reason
4136
Xid_apply_log_event::do_shall_skip(rpl_group_info *rgi)
4137
{
4138
  DBUG_ENTER("Xid_apply_log_event::do_shall_skip");
4139 4140 4141 4142 4143 4144 4145
  if (rgi->rli->slave_skip_counter > 0)
  {
    DBUG_ASSERT(!rgi->rli->get_flag(Relay_log_info::IN_TRANSACTION));
    thd->variables.option_bits&= ~(OPTION_BEGIN | OPTION_GTID_BEGIN);
    DBUG_RETURN(Log_event::EVENT_SKIP_COUNT);
  }
#ifdef WITH_WSREP
Marko Mäkelä's avatar
Marko Mäkelä committed
4146
  else if (wsrep_mysql_replication_bundle && WSREP(thd) &&
4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161
           opt_slave_domain_parallel_threads == 0)
  {
    if (++thd->wsrep_mysql_replicated < (int)wsrep_mysql_replication_bundle)
    {
      WSREP_DEBUG("skipping wsrep commit %d", thd->wsrep_mysql_replicated);
      DBUG_RETURN(Log_event::EVENT_SKIP_IGNORE);
    }
    else
    {
      thd->wsrep_mysql_replicated = 0;
    }
  }
#endif
  DBUG_RETURN(Log_event::do_shall_skip(rgi));
}
4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182
#endif /* HAVE_REPLICATION */

/**************************************************************************
  Xid_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void Xid_log_event::pack_info(Protocol *protocol)
{
  char buf[128], *pos;
  pos= strmov(buf, "COMMIT /* xid=");
  pos= longlong10_to_str(xid, pos, 10);
  pos= strmov(pos, " */");
  protocol->store(buf, (uint) (pos-buf), &my_charset_bin);
}


int Xid_log_event::do_commit()
{
  bool res;
  res= trans_commit(thd); /* Automatically rolls back on error. */
Marko Mäkelä's avatar
Marko Mäkelä committed
4183
  thd->release_transactional_locks();
4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224
  return res;
}
#endif


bool Xid_log_event::write()
{
  DBUG_EXECUTE_IF("do_not_write_xid", return 0;);
  return write_header(sizeof(xid)) ||
         write_data((uchar*)&xid, sizeof(xid)) ||
         write_footer();
}

/**************************************************************************
  XA_prepare_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
void XA_prepare_log_event::pack_info(Protocol *protocol)
{
  char query[sizeof("XA COMMIT ONE PHASE") + 1 + ser_buf_size];

  sprintf(query,
          (one_phase ? "XA COMMIT %s ONE PHASE" :  "XA PREPARE %s"),
          m_xid.serialize());

  protocol->store(query, strlen(query), &my_charset_bin);
}


int XA_prepare_log_event::do_commit()
{
  int res;
  xid_t xid;
  xid.set(m_xid.formatID,
          m_xid.data, m_xid.gtrid_length,
          m_xid.data + m_xid.gtrid_length, m_xid.bqual_length);

  thd->lex->xid= &xid;
  if (!one_phase)
  {
4225 4226
    if (thd->is_current_stmt_binlog_disabled() &&
        (res= thd->wait_for_prior_commit()))
4227 4228 4229 4230 4231 4232 4233 4234 4235 4236
      return res;

    thd->lex->sql_command= SQLCOM_XA_PREPARE;
    res= trans_xa_prepare(thd);
  }
  else
    res= trans_xa_commit(thd);

  return res;
}
4237 4238 4239
#endif // HAVE_REPLICATION


4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261
bool XA_prepare_log_event::write()
{
  uchar data[1 + 4 + 4 + 4]= {one_phase,};
  uint8 one_phase_byte= one_phase;

  int4store(data+1, static_cast<XID*>(xid)->formatID);
  int4store(data+(1+4), static_cast<XID*>(xid)->gtrid_length);
  int4store(data+(1+4+4), static_cast<XID*>(xid)->bqual_length);

  DBUG_ASSERT(xid_subheader_no_data == sizeof(data) - 1);

  return write_header(sizeof(one_phase_byte) + xid_subheader_no_data +
                      static_cast<XID*>(xid)->gtrid_length +
                      static_cast<XID*>(xid)->bqual_length) ||
         write_data(data, sizeof(data)) ||
         write_data((uchar*) static_cast<XID*>(xid)->data,
                     static_cast<XID*>(xid)->gtrid_length +
                     static_cast<XID*>(xid)->bqual_length) ||
         write_footer();
}


4262 4263 4264 4265 4266 4267 4268
/**************************************************************************
  User_var_log_event methods
**************************************************************************/

#if defined(HAVE_REPLICATION)
static bool
user_var_append_name_part(THD *thd, String *buf,
4269 4270
                          const char *name, size_t name_len,
                          const LEX_CSTRING &data_type_name)
4271
{
Monty's avatar
Monty committed
4272
  return buf->append('@') ||
4273
    append_identifier(thd, buf, name, name_len) ||
Marko Mäkelä's avatar
Marko Mäkelä committed
4274
    buf->append('=') ||
4275
    (data_type_name.length &&
Marko Mäkelä's avatar
Marko Mäkelä committed
4276
     (buf->append(STRING_WITH_LEN("/*")) ||
4277
      buf->append(data_type_name.str, data_type_name.length) ||
Marko Mäkelä's avatar
Marko Mäkelä committed
4278
      buf->append(STRING_WITH_LEN("*/"))));
4279 4280 4281 4282 4283 4284 4285 4286 4287
}

void User_var_log_event::pack_info(Protocol* protocol)
{
  if (is_null)
  {
    char buf_mem[FN_REFLEN+7];
    String buf(buf_mem, sizeof(buf_mem), system_charset_info);
    buf.length(0);
4288 4289
    if (user_var_append_name_part(protocol->thd, &buf, name, name_len,
                                  m_data_type_name) ||
Monty's avatar
Monty committed
4290
        buf.append(NULL_clex_str))
4291 4292 4293 4294 4295
      return;
    protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
  }
  else
  {
4296
    switch (m_type) {
4297 4298 4299 4300 4301 4302 4303 4304
    case REAL_RESULT:
    {
      double real_val;
      char buf2[MY_GCVT_MAX_FIELD_WIDTH+1];
      char buf_mem[FN_REFLEN + MY_GCVT_MAX_FIELD_WIDTH + 1];
      String buf(buf_mem, sizeof(buf_mem), system_charset_info);
      float8get(real_val, val);
      buf.length(0);
4305 4306
      if (user_var_append_name_part(protocol->thd, &buf, name, name_len,
                                    m_data_type_name) ||
4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318
          buf.append(buf2, my_gcvt(real_val, MY_GCVT_ARG_DOUBLE,
                                   MY_GCVT_MAX_FIELD_WIDTH, buf2, NULL)))
        return;
      protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
      break;
    }
    case INT_RESULT:
    {
      char buf2[22];
      char buf_mem[FN_REFLEN + 22];
      String buf(buf_mem, sizeof(buf_mem), system_charset_info);
      buf.length(0);
4319 4320
      if (user_var_append_name_part(protocol->thd, &buf, name, name_len,
                                    m_data_type_name) ||
4321 4322
          buf.append(buf2,
                 longlong10_to_str(uint8korr(val), buf2,
4323
                   (is_unsigned() ? 10 : -10))-buf2))
4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335
        return;
      protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
      break;
    }
    case DECIMAL_RESULT:
    {
      char buf_mem[FN_REFLEN + DECIMAL_MAX_STR_LENGTH];
      String buf(buf_mem, sizeof(buf_mem), system_charset_info);
      char buf2[DECIMAL_MAX_STR_LENGTH+1];
      String str(buf2, sizeof(buf2), &my_charset_bin);
      buf.length(0);
      my_decimal((const uchar *) (val + 2), val[0], val[1]).to_string(&str);
4336 4337
      if (user_var_append_name_part(protocol->thd, &buf, name, name_len,
                                    m_data_type_name) ||
Monty's avatar
Monty committed
4338
          buf.append(str))
4339 4340
        return;
      protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
Monty's avatar
Monty committed
4341

4342 4343 4344 4345 4346 4347 4348 4349 4350
      break;
    }
    case STRING_RESULT:
    {
      /* 15 is for 'COLLATE' and other chars */
      char buf_mem[FN_REFLEN + 512 + 1 + 2*MY_CS_NAME_SIZE+15];
      String buf(buf_mem, sizeof(buf_mem), system_charset_info);
      CHARSET_INFO *cs;
      buf.length(0);
4351
      if (!(cs= get_charset(m_charset_number, MYF(0))))
4352
      {
Monty's avatar
Monty committed
4353
        if (buf.append(STRING_WITH_LEN("???")))
4354 4355 4356 4357 4358 4359
          return;
      }
      else
      {
        size_t old_len;
        char *beg, *end;
4360 4361
        if (user_var_append_name_part(protocol->thd, &buf, name, name_len,
                                      m_data_type_name) ||
Monty's avatar
Monty committed
4362
            buf.append('_') ||
4363
            buf.append(cs->cs_name) ||
Monty's avatar
Monty committed
4364
            buf.append(' '))
4365 4366 4367 4368 4369 4370 4371 4372
          return;
        old_len= buf.length();
        if (buf.reserve(old_len + val_len * 2 + 3 + sizeof(" COLLATE ") +
                        MY_CS_NAME_SIZE))
          return;
        beg= const_cast<char *>(buf.ptr()) + old_len;
        end= str_to_hex(beg, val, val_len);
        buf.length(old_len + (end - beg));
Monty's avatar
Monty committed
4373
        if (buf.append(STRING_WITH_LEN(" COLLATE ")) ||
4374
            buf.append(cs->coll_name))
4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408
          return;
      }
      protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
      break;
    }
    case ROW_RESULT:
    default:
      DBUG_ASSERT(0);
      return;
    }
  }
}
#endif // HAVE_REPLICATION


bool User_var_log_event::write()
{
  char buf[UV_NAME_LEN_SIZE];
  char buf1[UV_VAL_IS_NULL + UV_VAL_TYPE_SIZE + 
	    UV_CHARSET_NUMBER_SIZE + UV_VAL_LEN_SIZE];
  uchar buf2[MY_MAX(8, DECIMAL_MAX_FIELD_SIZE + 2)], *pos= buf2;
  uint unsigned_len= 0;
  uint buf1_length;
  size_t event_length;

  int4store(buf, name_len);
  
  if ((buf1[0]= is_null))
  {
    buf1_length= 1;
    val_len= 0;                                 // Length of 'pos'
  }    
  else
  {
4409 4410
    buf1[1]= m_type;
    int4store(buf1 + 2, m_charset_number);
4411

4412
    switch (m_type) {
4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441
    case REAL_RESULT:
      float8store(buf2, *(double*) val);
      break;
    case INT_RESULT:
      int8store(buf2, *(longlong*) val);
      unsigned_len= 1;
      break;
    case DECIMAL_RESULT:
    {
      my_decimal *dec= (my_decimal *)val;
      dec->fix_buffer_pointer();
      buf2[0]= (char)(dec->intg + dec->frac);
      buf2[1]= (char)dec->frac;
      decimal2bin((decimal_t*)val, buf2+2, buf2[0], buf2[1]);
      val_len= decimal_bin_size(buf2[0], buf2[1]) + 2;
      break;
    }
    case STRING_RESULT:
      pos= (uchar*) val;
      break;
    case ROW_RESULT:
    default:
      DBUG_ASSERT(0);
      return 0;
    }
    int4store(buf1 + 2 + UV_CHARSET_NUMBER_SIZE, val_len);
    buf1_length= 10;
  }

4442 4443 4444 4445
  uchar data_type_name_chunk_signature= (uchar) CHUNK_DATA_TYPE_NAME;
  uint data_type_name_chunk_signature_length= m_data_type_name.length ? 1 : 0;
  uchar data_type_name_length_length= m_data_type_name.length ? 1 : 0;

4446
  /* Length of the whole event */
4447 4448 4449 4450
  event_length= sizeof(buf)+ name_len + buf1_length + val_len + unsigned_len +
                data_type_name_chunk_signature_length +
                data_type_name_length_length +
                (uint) m_data_type_name.length;
4451

4452 4453
  uchar unsig= m_is_unsigned ? CHUNK_UNSIGNED : CHUNK_SIGNED;
  uchar data_type_name_length= (uchar) m_data_type_name.length;
4454 4455 4456 4457 4458
  return write_header(event_length) ||
         write_data(buf, sizeof(buf))   ||
         write_data(name, name_len)     ||
         write_data(buf1, buf1_length) ||
         write_data(pos, val_len) ||
4459 4460 4461 4462 4463
         write_data(&unsig, unsigned_len) ||
         write_data(&data_type_name_chunk_signature,
                    data_type_name_chunk_signature_length) ||
         write_data(&data_type_name_length, data_type_name_length_length) ||
         write_data(m_data_type_name.str, (uint) m_data_type_name.length) ||
4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486
         write_footer();
}


#if defined(HAVE_REPLICATION)
int User_var_log_event::do_apply_event(rpl_group_info *rgi)
{
  Item *it= 0;
  CHARSET_INFO *charset;
  DBUG_ENTER("User_var_log_event::do_apply_event");
  query_id_t sav_query_id= 0; /* memorize orig id when deferred applying */

  if (rgi->deferred_events_collecting)
  {
    set_deferred(current_thd->query_id);
    DBUG_RETURN(rgi->deferred_events->add(this));
  }
  else if (is_deferred())
  {
    sav_query_id= current_thd->query_id;
    current_thd->query_id= query_id; /* recreating original time context */
  }

4487
  if (!(charset= get_charset(m_charset_number, MYF(MY_WME))))
4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505
  {
    rgi->rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                "Invalid character set for User var event");
    DBUG_RETURN(1);
  }
  LEX_CSTRING user_var_name;
  user_var_name.str= name;
  user_var_name.length= name_len;
  double real_val;
  longlong int_val;

  if (is_null)
  {
    it= new (thd->mem_root) Item_null(thd);
  }
  else
  {
4506
    switch (m_type) {
4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569
    case REAL_RESULT:
      if (val_len != 8)
      {
        rgi->rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                    ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                    "Invalid variable length at User var event");
        return 1;
      }
      float8get(real_val, val);
      it= new (thd->mem_root) Item_float(thd, real_val, 0);
      val= (char*) &real_val;		// Pointer to value in native format
      val_len= 8;
      break;
    case INT_RESULT:
      if (val_len != 8)
      {
        rgi->rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                    ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                    "Invalid variable length at User var event");
        return 1;
      }
      int_val= (longlong) uint8korr(val);
      it= new (thd->mem_root) Item_int(thd, int_val);
      val= (char*) &int_val;		// Pointer to value in native format
      val_len= 8;
      break;
    case DECIMAL_RESULT:
    {
      if (val_len < 3)
      {
        rgi->rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
                    ER_THD(thd, ER_SLAVE_FATAL_ERROR),
                    "Invalid variable length at User var event");
        return 1;
      }
      Item_decimal *dec= new (thd->mem_root) Item_decimal(thd, (uchar*) val+2, val[0], val[1]);
      it= dec;
      val= (char *)dec->val_decimal(NULL);
      val_len= sizeof(my_decimal);
      break;
    }
    case STRING_RESULT:
      it= new (thd->mem_root) Item_string(thd, val, (uint)val_len, charset);
      break;
    case ROW_RESULT:
    default:
      DBUG_ASSERT(0);
      DBUG_RETURN(0);
    }
  }

  Item_func_set_user_var *e= new (thd->mem_root) Item_func_set_user_var(thd, &user_var_name, it);
  /*
    Item_func_set_user_var can't substitute something else on its place =>
    0 can be passed as last argument (reference on item)

    Fix_fields() can fail, in which case a call of update_hash() might
    crash the server, so if fix fields fails, we just return with an
    error.
  */
  if (e->fix_fields(thd, 0))
    DBUG_RETURN(1);

4570 4571 4572 4573
  const Type_handler *th= Type_handler::handler_by_log_event_data_type(thd,
                                                                       *this);
  e->update_hash((void*) val, val_len, th, charset);

4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072
  if (!is_deferred())
    free_root(thd->mem_root, 0);
  else
    current_thd->query_id= sav_query_id; /* restore current query's context */

  DBUG_RETURN(0);
}

int User_var_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}

Log_event::enum_skip_reason
User_var_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    It is a common error to set the slave skip counter to 1 instead
    of 2 when recovering from an insert which used a auto increment,
    rand, or user var.  Therefore, if the slave skip counter is 1, we
    just say that this event should be skipped by ignoring it, meaning
    that we do not change the value of the slave skip counter since it
    will be decreased by the following insert event.
  */
  return continue_group(rgi);
}
#endif // HAVE_REPLICATION


#ifdef HAVE_REPLICATION

/**************************************************************************
	Stop_log_event methods
**************************************************************************/

/*
  The master stopped.  We used to clean up all temporary tables but
  this is useless as, as the master has shut down properly, it has
  written all DROP TEMPORARY TABLE (prepared statements' deletion is
  TODO only when we binlog prep stmts).  We used to clean up
  slave_load_tmpdir, but this is useless as it has been cleared at the
  end of LOAD DATA INFILE.  So we have nothing to do here.  The place
  were we must do this cleaning is in
  Start_log_event_v3::do_apply_event(), not here. Because if we come
  here, the master was sane.

  This must only be called from the Slave SQL thread, since it calls
  Relay_log_info::flush().
*/

int Stop_log_event::do_update_pos(rpl_group_info *rgi)
{
  int error= 0;
  Relay_log_info *rli= rgi->rli;
  DBUG_ENTER("Stop_log_event::do_update_pos");
  /*
    We do not want to update master_log pos because we get a rotate event
    before stop, so by now group_master_log_name is set to the next log.
    If we updated it, we will have incorrect master coordinates and this
    could give false triggers in MASTER_POS_WAIT() that we have reached
    the target position when in fact we have not.
  */
  if (rli->get_flag(Relay_log_info::IN_TRANSACTION))
    rgi->inc_event_relay_log_pos();
  else if (!rgi->is_parallel_exec)
  {
    rpl_global_gtid_slave_state->record_and_update_gtid(thd, rgi);
    rli->inc_group_relay_log_pos(0, rgi);
    if (rli->flush())
      error= 1;
  }
  DBUG_RETURN(error);
}

#endif /* HAVE_REPLICATION */


/**************************************************************************
	Create_file_log_event methods
**************************************************************************/

Create_file_log_event::
Create_file_log_event(THD* thd_arg, sql_exchange* ex,
		      const char* db_arg, const char* table_name_arg,
                      List<Item>& fields_arg,
                      bool is_concurrent_arg,
                      enum enum_duplicates handle_dup,
                      bool ignore,
		      uchar* block_arg, uint block_len_arg, bool using_trans)
  :Load_log_event(thd_arg, ex, db_arg, table_name_arg, fields_arg,
                  is_concurrent_arg,
                  handle_dup, ignore, using_trans),
   fake_base(0), block(block_arg), event_buf(0), block_len(block_len_arg),
   file_id(thd_arg->file_id = mysql_bin_log.next_file_id())
{
  DBUG_ENTER("Create_file_log_event");
  sql_ex.force_new_format();
  DBUG_VOID_RETURN;
}


/*
  Create_file_log_event::write_data_body()
*/

bool Create_file_log_event::write_data_body()
{
  bool res;
  if ((res= Load_log_event::write_data_body()) || fake_base)
    return res;
  return write_data("", 1) ||
         write_data(block, block_len);
}


/*
  Create_file_log_event::write_data_header()
*/

bool Create_file_log_event::write_data_header()
{
  bool res;
  uchar buf[CREATE_FILE_HEADER_LEN];
  if ((res= Load_log_event::write_data_header()) || fake_base)
    return res;
  int4store(buf + CF_FILE_ID_OFFSET, file_id);
  return write_data(buf, CREATE_FILE_HEADER_LEN) != 0;
}


/*
  Create_file_log_event::write_base()
*/

bool Create_file_log_event::write_base()
{
  bool res;
  fake_base= 1;                                 // pretend we are Load event
  res= write();
  fake_base= 0;
  return res;
}


#if defined(HAVE_REPLICATION)
void Create_file_log_event::pack_info(Protocol *protocol)
{
  char buf[SAFE_NAME_LEN*2 + 30 + 21*2], *pos;
  pos= strmov(buf, "db=");
  memcpy(pos, db, db_len);
  pos= strmov(pos + db_len, ";table=");
  memcpy(pos, table_name, table_name_len);
  pos= strmov(pos + table_name_len, ";file_id=");
  pos= int10_to_str((long) file_id, pos, 10);
  pos= strmov(pos, ";block_len=");
  pos= int10_to_str((long) block_len, pos, 10);
  protocol->store(buf, (uint) (pos-buf), &my_charset_bin);
}
#endif /* defined(HAVE_REPLICATION) */


/**
  Create_file_log_event::do_apply_event()
  Constructor for Create_file_log_event to intantiate an event
  from the relay log on the slave.

  @retval
    0           Success
  @retval
    1           Failure
*/

#if defined(HAVE_REPLICATION)
int Create_file_log_event::do_apply_event(rpl_group_info *rgi)
{
  char fname_buf[FN_REFLEN];
  char *ext;
  int fd = -1;
  IO_CACHE file;
  Log_event_writer lew(&file, 0);
  int error = 1;
  Relay_log_info const *rli= rgi->rli;

  THD_STAGE_INFO(thd, stage_making_temp_file_create_before_load_data);
  bzero((char*)&file, sizeof(file));
  ext= slave_load_file_stem(fname_buf, file_id, server_id, ".info",
                            &rli->mi->connection_name);
  /* old copy may exist already */
  mysql_file_delete(key_file_log_event_info, fname_buf, MYF(0));
  if ((fd= mysql_file_create(key_file_log_event_info,
                             fname_buf, CREATE_MODE,
                             O_WRONLY | O_BINARY | O_EXCL | O_NOFOLLOW,
                             MYF(MY_WME))) < 0 ||
      init_io_cache(&file, fd, IO_SIZE, WRITE_CACHE, (my_off_t)0, 0,
		    MYF(MY_WME|MY_NABP)))
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in Create_file event: could not open file '%s'",
                fname_buf);
    goto err;
  }

  // a trick to avoid allocating another buffer
  fname= fname_buf;
  fname_len= (uint) (strmov(ext, ".data") - fname);
  writer= &lew;
  if (write_base())
  {
    strmov(ext, ".info"); // to have it right in the error message
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in Create_file event: could not write to file '%s'",
                fname_buf);
    goto err;
  }
  end_io_cache(&file);
  mysql_file_close(fd, MYF(0));

  // fname_buf now already has .data, not .info, because we did our trick
  /* old copy may exist already */
  mysql_file_delete(key_file_log_event_data, fname_buf, MYF(0));
  if ((fd= mysql_file_create(key_file_log_event_data,
                             fname_buf, CREATE_MODE,
                             O_WRONLY | O_BINARY | O_EXCL | O_NOFOLLOW,
                             MYF(MY_WME))) < 0)
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in Create_file event: could not open file '%s'",
                fname_buf);
    goto err;
  }
  if (mysql_file_write(fd, (uchar*) block, block_len, MYF(MY_WME+MY_NABP)))
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in Create_file event: write to '%s' failed",
                fname_buf);
    goto err;
  }
  error=0;					// Everything is ok

err:
  if (unlikely(error))
    end_io_cache(&file);
  if (likely(fd >= 0))
    mysql_file_close(fd, MYF(0));
  return error != 0;
}
#endif /* defined(HAVE_REPLICATION) */


/**************************************************************************
	Append_block_log_event methods
**************************************************************************/

Append_block_log_event::Append_block_log_event(THD *thd_arg,
                                               const char *db_arg,
					       uchar *block_arg,
					       uint block_len_arg,
					       bool using_trans)
  :Log_event(thd_arg,0, using_trans), block(block_arg),
   block_len(block_len_arg), file_id(thd_arg->file_id), db(db_arg)
{
}


bool Append_block_log_event::write()
{
  uchar buf[APPEND_BLOCK_HEADER_LEN];
  int4store(buf + AB_FILE_ID_OFFSET, file_id);
  return write_header(APPEND_BLOCK_HEADER_LEN + block_len) ||
         write_data(buf, APPEND_BLOCK_HEADER_LEN) ||
         write_data(block, block_len) ||
         write_footer();
}


#if defined(HAVE_REPLICATION)
void Append_block_log_event::pack_info(Protocol *protocol)
{
  char buf[256];
  uint length;
  length= (uint) sprintf(buf, ";file_id=%u;block_len=%u", file_id, block_len);
  protocol->store(buf, length, &my_charset_bin);
}


/*
  Append_block_log_event::get_create_or_append()
*/

int Append_block_log_event::get_create_or_append() const
{
  return 0; /* append to the file, fail if not exists */
}

/*
  Append_block_log_event::do_apply_event()
*/

int Append_block_log_event::do_apply_event(rpl_group_info *rgi)
{
  char fname[FN_REFLEN];
  int fd;
  int error = 1;
  Relay_log_info const *rli= rgi->rli;
  DBUG_ENTER("Append_block_log_event::do_apply_event");

  THD_STAGE_INFO(thd, stage_making_temp_file_append_before_load_data);
  slave_load_file_stem(fname, file_id, server_id, ".data",
                       &rli->mi->cmp_connection_name);
  if (get_create_or_append())
  {
    /*
      Usually lex_start() is called by mysql_parse(), but we need it here
      as the present method does not call mysql_parse().
    */
    lex_start(thd);
    thd->reset_for_next_command();
    /* old copy may exist already */
    mysql_file_delete(key_file_log_event_data, fname, MYF(0));
    if ((fd= mysql_file_create(key_file_log_event_data,
                               fname, CREATE_MODE,
                               O_WRONLY | O_BINARY | O_EXCL | O_NOFOLLOW,
                               MYF(MY_WME))) < 0)
    {
      rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                  "Error in %s event: could not create file '%s'",
                  get_type_str(), fname);
      goto err;
    }
  }
  else if ((fd= mysql_file_open(key_file_log_event_data,
                                fname,
                                O_WRONLY | O_APPEND | O_BINARY | O_NOFOLLOW,
                                MYF(MY_WME))) < 0)
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in %s event: could not open file '%s'",
                get_type_str(), fname);
    goto err;
  }

  DBUG_EXECUTE_IF("remove_slave_load_file_before_write",
                  {
                    my_delete(fname, MYF(0));
                  });

  if (mysql_file_write(fd, (uchar*) block, block_len, MYF(MY_WME+MY_NABP)))
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in %s event: write to '%s' failed",
                get_type_str(), fname);
    goto err;
  }
  error=0;

err:
  if (fd >= 0)
    mysql_file_close(fd, MYF(0));
  DBUG_RETURN(error);
}
#endif // HAVE_REPLICATION


/**************************************************************************
	Delete_file_log_event methods
**************************************************************************/

Delete_file_log_event::Delete_file_log_event(THD *thd_arg, const char* db_arg,
					     bool using_trans)
  :Log_event(thd_arg, 0, using_trans), file_id(thd_arg->file_id), db(db_arg)
{
}


bool Delete_file_log_event::write()
{
 uchar buf[DELETE_FILE_HEADER_LEN];
 int4store(buf + DF_FILE_ID_OFFSET, file_id);
 return write_header(sizeof(buf)) ||
        write_data(buf, sizeof(buf)) ||
        write_footer();
}


#if defined(HAVE_REPLICATION)
void Delete_file_log_event::pack_info(Protocol *protocol)
{
  char buf[64];
  uint length;
  length= (uint) sprintf(buf, ";file_id=%u", (uint) file_id);
  protocol->store(buf, (int32) length, &my_charset_bin);
}
#endif


#if defined(HAVE_REPLICATION)
int Delete_file_log_event::do_apply_event(rpl_group_info *rgi)
{
  char fname[FN_REFLEN+10];
  Relay_log_info const *rli= rgi->rli;
  char *ext= slave_load_file_stem(fname, file_id, server_id, ".data",
                                  &rli->mi->cmp_connection_name);
  mysql_file_delete(key_file_log_event_data, fname, MYF(MY_WME));
  strmov(ext, ".info");
  mysql_file_delete(key_file_log_event_info, fname, MYF(MY_WME));
  return 0;
}
#endif /* defined(HAVE_REPLICATION) */


/**************************************************************************
	Execute_load_log_event methods
**************************************************************************/

Execute_load_log_event::Execute_load_log_event(THD *thd_arg,
                                               const char* db_arg,
					       bool using_trans)
  :Log_event(thd_arg, 0, using_trans), file_id(thd_arg->file_id), db(db_arg)
{
}
  

bool Execute_load_log_event::write()
{
  uchar buf[EXEC_LOAD_HEADER_LEN];
  int4store(buf + EL_FILE_ID_OFFSET, file_id);
  return write_header(sizeof(buf)) ||
         write_data(buf, sizeof(buf)) ||
         write_footer();
}


#if defined(HAVE_REPLICATION)
void Execute_load_log_event::pack_info(Protocol *protocol)
{
  char buf[64];
  uint length;
  length= (uint) sprintf(buf, ";file_id=%u", (uint) file_id);
  protocol->store(buf, (int32) length, &my_charset_bin);
}


/*
  Execute_load_log_event::do_apply_event()
*/

int Execute_load_log_event::do_apply_event(rpl_group_info *rgi)
{
  char fname[FN_REFLEN+10];
  char *ext;
  int fd;
  int error= 1;
  IO_CACHE file;
  Load_log_event *lev= 0;
  Relay_log_info const *rli= rgi->rli;

  ext= slave_load_file_stem(fname, file_id, server_id, ".info",
                            &rli->mi->cmp_connection_name);
  if ((fd= mysql_file_open(key_file_log_event_info,
                           fname, O_RDONLY | O_BINARY | O_NOFOLLOW,
                           MYF(MY_WME))) < 0 ||
      init_io_cache(&file, fd, IO_SIZE, READ_CACHE, (my_off_t)0, 0,
		    MYF(MY_WME|MY_NABP)))
  {
    rli->report(ERROR_LEVEL, my_errno, rgi->gtid_info(),
                "Error in Exec_load event: could not open file '%s'",
                fname);
    goto err;
  }
  if (!(lev= (Load_log_event*)
        Log_event::read_log_event(&file,
                                  rli->relay_log.description_event_for_exec,
                                  opt_slave_sql_verify_checksum)) ||
      lev->get_type_code() != NEW_LOAD_EVENT)
  {
    rli->report(ERROR_LEVEL, 0, rgi->gtid_info(), "Error in Exec_load event: "
                    "file '%s' appears corrupted", fname);
    goto err;
  }
  lev->thd = thd;
  /*
    lev->do_apply_event should use rli only for errors i.e. should
    not advance rli's position.

    lev->do_apply_event is the place where the table is loaded (it
    calls mysql_load()).
  */

  if (lev->do_apply_event(0,rgi,1)) 
  {
    /*
      We want to indicate the name of the file that could not be loaded
      (SQL_LOADxxx).
      But as we are here we are sure the error is in rli->last_slave_error and
      rli->last_slave_errno (example of error: duplicate entry for key), so we
      don't want to overwrite it with the filename.
      What we want instead is add the filename to the current error message.
    */
5073
    char *tmp= my_strdup(PSI_INSTRUMENT_ME, rli->last_error().message, MYF(MY_WME));
5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187
    if (tmp)
    {
      rli->report(ERROR_LEVEL, rli->last_error().number, rgi->gtid_info(),
                  "%s. Failed executing load from '%s'", tmp, fname);
      my_free(tmp);
    }
    goto err;
  }
  /*
    We have an open file descriptor to the .info file; we need to close it
    or Windows will refuse to delete the file in mysql_file_delete().
  */
  if (fd >= 0)
  {
    mysql_file_close(fd, MYF(0));
    end_io_cache(&file);
    fd= -1;
  }
  mysql_file_delete(key_file_log_event_info, fname, MYF(MY_WME));
  memcpy(ext, ".data", 6);
  mysql_file_delete(key_file_log_event_data, fname, MYF(MY_WME));
  error = 0;

err:
  delete lev;
  if (fd >= 0)
  {
    mysql_file_close(fd, MYF(0));
    end_io_cache(&file);
  }
  return error;
}

#endif /* defined(HAVE_REPLICATION) */

/**************************************************************************
	Begin_load_query_log_event methods
**************************************************************************/

Begin_load_query_log_event::
Begin_load_query_log_event(THD* thd_arg, const char* db_arg, uchar* block_arg,
                           uint block_len_arg, bool using_trans)
  :Append_block_log_event(thd_arg, db_arg, block_arg, block_len_arg,
                          using_trans)
{
   file_id= thd_arg->file_id= mysql_bin_log.next_file_id();
}


#if defined( HAVE_REPLICATION)
int Begin_load_query_log_event::get_create_or_append() const
{
  return 1; /* create the file */
}


Log_event::enum_skip_reason
Begin_load_query_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    If the slave skip counter is 1, then we should not start executing
    on the next event.
  */
  return continue_group(rgi);
}
#endif /* defined( HAVE_REPLICATION) */


/**************************************************************************
	Execute_load_query_log_event methods
**************************************************************************/

Execute_load_query_log_event::
Execute_load_query_log_event(THD *thd_arg, const char* query_arg,
                             ulong query_length_arg, uint fn_pos_start_arg,
                             uint fn_pos_end_arg,
                             enum_load_dup_handling dup_handling_arg,
                             bool using_trans, bool direct, bool suppress_use,
                             int errcode):
  Query_log_event(thd_arg, query_arg, query_length_arg, using_trans, direct,
                  suppress_use, errcode),
  file_id(thd_arg->file_id), fn_pos_start(fn_pos_start_arg),
  fn_pos_end(fn_pos_end_arg), dup_handling(dup_handling_arg)
{
}


bool
Execute_load_query_log_event::write_post_header_for_derived()
{
  uchar buf[EXECUTE_LOAD_QUERY_EXTRA_HEADER_LEN];
  int4store(buf, file_id);
  int4store(buf + 4, fn_pos_start);
  int4store(buf + 4 + 4, fn_pos_end);
  *(buf + 4 + 4 + 4)= (uchar) dup_handling;
  return write_data(buf, EXECUTE_LOAD_QUERY_EXTRA_HEADER_LEN);
}


#if defined(HAVE_REPLICATION)
void Execute_load_query_log_event::pack_info(Protocol *protocol)
{
  char buf_mem[1024];
  String buf(buf_mem, sizeof(buf_mem), system_charset_info);
  buf.real_alloc(9 + db_len + q_len + 10 + 21);
  if (db && db_len)
  {
    if (buf.append(STRING_WITH_LEN("use ")) ||
        append_identifier(protocol->thd, &buf, db, db_len) ||
        buf.append(STRING_WITH_LEN("; ")))
      return;
  }
  if (query && q_len && buf.append(query, q_len))
    return;
Monty's avatar
Monty committed
5188
  if (buf.append(STRING_WITH_LEN(" ;file_id=")) ||
5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204
      buf.append_ulonglong(file_id))
    return;
  protocol->store(buf.ptr(), buf.length(), &my_charset_bin);
}


int
Execute_load_query_log_event::do_apply_event(rpl_group_info *rgi)
{
  char *p;
  char *buf;
  char *fname;
  char *fname_end;
  int error;
  Relay_log_info const *rli= rgi->rli;

5205 5206
  buf= (char*) my_malloc(PSI_INSTRUMENT_ME, q_len + 1 -
     (fn_pos_end - fn_pos_start) + (FN_REFLEN + 10) + 10 + 8 + 5, MYF(MY_WME));
5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300

  DBUG_EXECUTE_IF("LOAD_DATA_INFILE_has_fatal_error", my_free(buf); buf= NULL;);

  /* Replace filename and LOCAL keyword in query before executing it */
  if (buf == NULL)
  {
    rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR, rgi->gtid_info(),
                ER_THD(rgi->thd, ER_SLAVE_FATAL_ERROR), "Not enough memory");
    return 1;
  }

  p= buf;
  memcpy(p, query, fn_pos_start);
  p+= fn_pos_start;
  fname= (p= strmake(p, STRING_WITH_LEN(" INFILE \'")));
  p= slave_load_file_stem(p, file_id, server_id, ".data",
                          &rli->mi->cmp_connection_name);
  fname_end= p= strend(p);                      // Safer than p=p+5
  *(p++)='\'';
  switch (dup_handling) {
  case LOAD_DUP_IGNORE:
    p= strmake(p, STRING_WITH_LEN(" IGNORE"));
    break;
  case LOAD_DUP_REPLACE:
    p= strmake(p, STRING_WITH_LEN(" REPLACE"));
    break;
  default:
    /* Ordinary load data */
    break;
  }
  p= strmake(p, STRING_WITH_LEN(" INTO "));
  p= strmake(p, query+fn_pos_end, q_len-fn_pos_end);

  error= Query_log_event::do_apply_event(rgi, buf, (uint32)(p-buf));

  /* Forging file name for deletion in same buffer */
  *fname_end= 0;

  /*
    If there was an error the slave is going to stop, leave the
    file so that we can re-execute this event at START SLAVE.
  */
  if (unlikely(!error))
    mysql_file_delete(key_file_log_event_data, fname, MYF(MY_WME));

  my_free(buf);
  return error;
}
#endif // HAVE_REPLICATION


/**************************************************************************
	sql_ex_info methods
**************************************************************************/

static bool write_str(Log_event_writer *writer, const char *str, uint length)
{
  uchar tmp[1];
  tmp[0]= (uchar) length;
  return (writer->write_data(tmp, sizeof(tmp)) ||
	  writer->write_data((uchar*) str, length));
}

bool sql_ex_info::write_data(Log_event_writer *writer)
{
  if (new_format())
  {
    return write_str(writer, field_term, field_term_len) ||
	   write_str(writer, enclosed,   enclosed_len) ||
	   write_str(writer, line_term,  line_term_len) ||
	   write_str(writer, line_start, line_start_len) ||
	   write_str(writer, escaped,    escaped_len) ||
	   writer->write_data((uchar*) &opt_flags, 1);
  }
  else
  {
    uchar old_ex[7];
    old_ex[0]= *field_term;
    old_ex[1]= *enclosed;
    old_ex[2]= *line_term;
    old_ex[3]= *line_start;
    old_ex[4]= *escaped;
    old_ex[5]=  opt_flags;
    old_ex[6]=  empty_flags;
    return writer->write_data(old_ex, sizeof(old_ex));
  }
}



/**************************************************************************
	Rows_log_event member functions
**************************************************************************/

5301 5302
Rows_log_event::Rows_log_event(THD *thd_arg, TABLE *tbl_arg,
                               ulonglong table_id,
5303 5304 5305 5306 5307
                               MY_BITMAP const *cols, bool is_transactional,
                               Log_event_type event_type)
  : Log_event(thd_arg, 0, is_transactional),
    m_row_count(0),
    m_table(tbl_arg),
5308
    m_table_id(table_id),
5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319
    m_width(tbl_arg ? tbl_arg->s->fields : 1),
    m_rows_buf(0), m_rows_cur(0), m_rows_end(0), m_flags(0),
    m_type(event_type), m_extra_row_data(0)
#ifdef HAVE_REPLICATION
    , m_curr_row(NULL), m_curr_row_end(NULL),
    m_key(NULL), m_key_info(NULL), m_key_nr(0),
    master_had_triggers(0)
#endif
{
  /*
    We allow a special form of dummy event when the table, and cols
5320
    are null and the table id is UINT32_MAX.  This is a temporary
5321 5322 5323
    solution, to be able to terminate a started statement in the
    binary log: the extraneous events will be removed in the future.
   */
5324 5325 5326
  DBUG_ASSERT((tbl_arg && tbl_arg->s &&
               (table_id & MAX_TABLE_MAP_ID) != UINT32_MAX) ||
              (!tbl_arg && !cols && (table_id & MAX_TABLE_MAP_ID) == UINT32_MAX));
5327 5328 5329 5330 5331 5332 5333 5334 5335

  if (thd_arg->variables.option_bits & OPTION_NO_FOREIGN_KEY_CHECKS)
    set_flags(NO_FOREIGN_KEY_CHECKS_F);
  if (thd_arg->variables.option_bits & OPTION_RELAXED_UNIQUE_CHECKS)
    set_flags(RELAXED_UNIQUE_CHECKS_F);
  if (thd_arg->variables.option_bits & OPTION_NO_CHECK_CONSTRAINT_CHECKS)
    set_flags(NO_CHECK_CONSTRAINT_CHECKS_F);
  /* if my_bitmap_init fails, caught in is_valid() */
  if (likely(!my_bitmap_init(&m_cols,
Monty's avatar
Monty committed
5336 5337 5338
                             m_width <= sizeof(m_bitbuf)*8 ? m_bitbuf : NULL,
                             m_width,
                             false)))
5339 5340 5341
  {
    /* Cols can be zero if this is a dummy binrows event */
    if (likely(cols != NULL))
Monty's avatar
Monty committed
5342
      bitmap_copy(&m_cols, cols);
5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411
  }
}


int Rows_log_event::do_add_row_data(uchar *row_data, size_t length)
{
  /*
    When the table has a primary key, we would probably want, by default, to
    log only the primary key value instead of the entire "before image". This
    would save binlog space. TODO
  */
  DBUG_ENTER("Rows_log_event::do_add_row_data");
  DBUG_PRINT("enter", ("row_data:%p  length: %lu", row_data,
                       (ulong) length));

  /*
    If length is zero, there is nothing to write, so we just
    return. Note that this is not an optimization, since calling
    realloc() with size 0 means free().
   */
  if (length == 0)
  {
    m_row_count++;
    DBUG_RETURN(0);
  }

  /*
    Don't print debug messages when running valgrind since they can
    trigger false warnings.
   */
#ifndef HAVE_valgrind
  DBUG_DUMP("row_data", row_data, MY_MIN(length, 32));
#endif

  DBUG_ASSERT(m_rows_buf <= m_rows_cur);
  DBUG_ASSERT(!m_rows_buf || (m_rows_end && m_rows_buf < m_rows_end));
  DBUG_ASSERT(m_rows_cur <= m_rows_end);

  /* The cast will always work since m_rows_cur <= m_rows_end */
  if (static_cast<size_t>(m_rows_end - m_rows_cur) <= length)
  {
    size_t const block_size= 1024;
    size_t cur_size= m_rows_cur - m_rows_buf;
    DBUG_EXECUTE_IF("simulate_too_big_row_case1",
                     cur_size= UINT_MAX32 - (block_size * 10);
                     length= UINT_MAX32 - (block_size * 10););
    DBUG_EXECUTE_IF("simulate_too_big_row_case2",
                     cur_size= UINT_MAX32 - (block_size * 10);
                     length= block_size * 10;);
    DBUG_EXECUTE_IF("simulate_too_big_row_case3",
                     cur_size= block_size * 10;
                     length= UINT_MAX32 - (block_size * 10););
    DBUG_EXECUTE_IF("simulate_too_big_row_case4",
                     cur_size= UINT_MAX32 - (block_size * 10);
                     length= (block_size * 10) - block_size + 1;);
    size_t remaining_space= UINT_MAX32 - cur_size;
    /* Check that the new data fits within remaining space and we can add
       block_size without wrapping.
     */
    if (cur_size > UINT_MAX32 || length > remaining_space ||
        ((length + block_size) > remaining_space))
    {
      sql_print_error("The row data is greater than 4GB, which is too big to "
                      "write to the binary log.");
      DBUG_RETURN(ER_BINLOG_ROW_LOGGING_FAILED);
    }
    size_t const new_alloc= 
        block_size * ((cur_size + length + block_size - 1) / block_size);

5412 5413
    uchar* const new_buf= (uchar*)my_realloc(PSI_INSTRUMENT_ME, m_rows_buf,
                                    new_alloc, MYF(MY_ALLOW_ZERO_PTR|MY_WME));
5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447
    if (unlikely(!new_buf))
      DBUG_RETURN(HA_ERR_OUT_OF_MEM);

    /* If the memory moved, we need to move the pointers */
    if (new_buf != m_rows_buf)
    {
      m_rows_buf= new_buf;
      m_rows_cur= m_rows_buf + cur_size;
    }

    /*
       The end pointer should always be changed to point to the end of
       the allocated memory.
    */
    m_rows_end= m_rows_buf + new_alloc;
  }

  DBUG_ASSERT(m_rows_cur + length <= m_rows_end);
  memcpy(m_rows_cur, row_data, length);
  m_rows_cur+= length;
  m_row_count++;
  DBUG_RETURN(0);
}


#if defined(HAVE_REPLICATION)

/**
  Restores empty table list as it was before trigger processing.

  @note We have a lot of ASSERTS that check the lists when we close tables.
  There was the same problem with MERGE MYISAM tables and so here we try to
  go the same way.
*/
Marko Mäkelä's avatar
Marko Mäkelä committed
5448
inline void restore_empty_query_table_list(LEX *lex)
5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462
{
  if (lex->first_not_own_table())
      (*lex->first_not_own_table()->prev_global)= NULL;
  lex->query_tables= NULL;
  lex->query_tables_last= &lex->query_tables;
}


int Rows_log_event::do_apply_event(rpl_group_info *rgi)
{
  Relay_log_info const *rli= rgi->rli;
  TABLE* table;
  DBUG_ENTER("Rows_log_event::do_apply_event(Relay_log_info*)");
  int error= 0;
Marko Mäkelä's avatar
Marko Mäkelä committed
5463 5464
  LEX *lex= thd->lex;
  uint8 new_trg_event_map= get_trg_event_map();
5465
  /*
5466
    If m_table_id == UINT32_MAX, then we have a dummy event that does not
5467 5468 5469 5470
    contain any data.  In that case, we just remove all tables in the
    tables_to_lock list, close the thread tables, and return with
    success.
   */
5471
  if (m_table_id == UINT32_MAX)
5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515
  {
    /*
       This one is supposed to be set: just an extra check so that
       nothing strange has happened.
     */
    DBUG_ASSERT(get_flags(STMT_END_F));

    rgi->slave_close_thread_tables(thd);
    thd->clear_error();
    DBUG_RETURN(0);
  }

  /*
    'thd' has been set by exec_relay_log_event(), just before calling
    do_apply_event(). We still check here to prevent future coding
    errors.
  */
  DBUG_ASSERT(rgi->thd == thd);

  /*
    If there is no locks taken, this is the first binrow event seen
    after the table map events.  We should then lock all the tables
    used in the transaction and proceed with execution of the actual
    event.
  */
  if (!thd->lock)
  {
    /*
      Lock_tables() reads the contents of thd->lex, so they must be
      initialized.

      We also call the THD::reset_for_next_command(), since this
      is the logical start of the next "statement". Note that this
      call might reset the value of current_stmt_binlog_format, so
      we need to do any changes to that value after this function.
    */
    delete_explain_query(thd->lex);
    lex_start(thd);
    thd->reset_for_next_command();
    /*
      The current statement is just about to begin and 
      has not yet modified anything. Note, all.modified is reset
      by THD::reset_for_next_command().
    */
5516 5517
    thd->transaction->stmt.modified_non_trans_table= FALSE;
    thd->transaction->stmt.m_unsafe_rollback_flags&= ~THD_TRANS::DID_WAIT;
5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548
    /*
      This is a row injection, so we flag the "statement" as
      such. Note that this code is called both when the slave does row
      injections and when the BINLOG statement is used to do row
      injections.
    */
    thd->lex->set_stmt_row_injection();

    /*
      There are a few flags that are replicated with each row event.
      Make sure to set/clear them before executing the main body of
      the event.
    */
    if (get_flags(NO_FOREIGN_KEY_CHECKS_F))
        thd->variables.option_bits|= OPTION_NO_FOREIGN_KEY_CHECKS;
    else
        thd->variables.option_bits&= ~OPTION_NO_FOREIGN_KEY_CHECKS;

    if (get_flags(RELAXED_UNIQUE_CHECKS_F))
        thd->variables.option_bits|= OPTION_RELAXED_UNIQUE_CHECKS;
    else
        thd->variables.option_bits&= ~OPTION_RELAXED_UNIQUE_CHECKS;

    if (get_flags(NO_CHECK_CONSTRAINT_CHECKS_F))
      thd->variables.option_bits|= OPTION_NO_CHECK_CONSTRAINT_CHECKS;
    else
      thd->variables.option_bits&= ~OPTION_NO_CHECK_CONSTRAINT_CHECKS;

    /* A small test to verify that objects have consistent types */
    DBUG_ASSERT(sizeof(thd->variables.option_bits) == sizeof(OPTION_RELAXED_UNIQUE_CHECKS));

Marko Mäkelä's avatar
Marko Mäkelä committed
5549 5550 5551 5552 5553 5554
    DBUG_EXECUTE_IF("rows_log_event_before_open_table",
                    {
                      const char action[] = "now SIGNAL before_open_table WAIT_FOR go_ahead_sql";
                      DBUG_ASSERT(!debug_sync_set_action(thd, STRING_WITH_LEN(action)));
                    };);

5555

Marko Mäkelä's avatar
Marko Mäkelä committed
5556 5557 5558
    /*
      Trigger's procedures work with global table list. So we have to add
      rgi->tables_to_lock content there to get trigger's in the list.
5559

Marko Mäkelä's avatar
Marko Mäkelä committed
5560 5561 5562 5563 5564
      Then restore_empty_query_table_list() restore the list as it was
    */
    DBUG_ASSERT(lex->query_tables == NULL);
    if ((lex->query_tables= rgi->tables_to_lock))
      rgi->tables_to_lock->prev_global= &lex->query_tables;
5565

Marko Mäkelä's avatar
Marko Mäkelä committed
5566 5567 5568 5569
    for (TABLE_LIST *tables= rgi->tables_to_lock; tables;
         tables= tables->next_global)
    {
      if (slave_run_triggers_for_rbr)
5570 5571 5572 5573
      {
        tables->trg_event_map= new_trg_event_map;
        lex->query_tables_last= &tables->next_global;
      }
Marko Mäkelä's avatar
Marko Mäkelä committed
5574
      else
Marko Mäkelä's avatar
Marko Mäkelä committed
5575 5576 5577 5578
      {
        tables->slave_fk_event_map= new_trg_event_map;
        lex->query_tables_last= &tables->next_global;
      }
5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593
    }
    if (unlikely(open_and_lock_tables(thd, rgi->tables_to_lock, FALSE, 0)))
    {
#ifdef WITH_WSREP
      if (WSREP(thd))
      {
        WSREP_WARN("BF applier failed to open_and_lock_tables: %u, fatal: %d "
                   "wsrep = (exec_mode: %d conflict_state: %d seqno: %lld)",
                    thd->get_stmt_da()->sql_errno(),
                    thd->is_fatal_error,
                    thd->wsrep_cs().mode(),
                    thd->wsrep_trx().state(),
                    (long long) wsrep_thd_trx_seqno(thd));
      }
#endif /* WITH_WSREP */
Marko Mäkelä's avatar
Marko Mäkelä committed
5594 5595
      if (thd->is_error() &&
          !is_parallel_retry_error(rgi, error= thd->get_stmt_da()->sql_errno()))
5596 5597 5598
      {
        /*
          Error reporting borrowed from Query_log_event with many excessive
Marko Mäkelä's avatar
Marko Mäkelä committed
5599
          simplifications.
5600
          We should not honour --slave-skip-errors at this point as we are
Marko Mäkelä's avatar
Marko Mäkelä committed
5601
          having severe errors which should not be skipped.
5602
        */
Marko Mäkelä's avatar
Marko Mäkelä committed
5603
        rli->report(ERROR_LEVEL, error, rgi->gtid_info(),
5604
                    "Error executing row event: '%s'",
Marko Mäkelä's avatar
Marko Mäkelä committed
5605
                    (error ? thd->get_stmt_da()->message() :
5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618
                     "unexpected success or fatal error"));
        thd->is_slave_error= 1;
      }
      /* remove trigger's tables */
      goto err;
    }

    /*
      When the open and locking succeeded, we check all tables to
      ensure that they still have the correct type.
    */

    {
5619
      DBUG_PRINT("debug", ("Checking compatibility of tables to lock - tables_to_lock: %p",
5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725
                           rgi->tables_to_lock));

      /**
        When using RBR and MyISAM MERGE tables the base tables that make
        up the MERGE table can be appended to the list of tables to lock.
  
        Thus, we just check compatibility for those that tables that have
        a correspondent table map event (ie, those that are actually going
        to be accessed while applying the event). That's why the loop stops
        at rli->tables_to_lock_count .

        NOTE: The base tables are added here are removed when 
              close_thread_tables is called.
       */
      TABLE_LIST *table_list_ptr= rgi->tables_to_lock;
      for (uint i=0 ; table_list_ptr && (i < rgi->tables_to_lock_count);
           table_list_ptr= table_list_ptr->next_global, i++)
      {
        /*
          Below if condition takes care of skipping base tables that
          make up the MERGE table (which are added by open_tables()
          call). They are added next to the merge table in the list.
          For eg: If RPL_TABLE_LIST is t3->t1->t2 (where t1 and t2
          are base tables for merge table 't3'), open_tables will modify
          the list by adding t1 and t2 again immediately after t3 in the
          list (*not at the end of the list*). New table_to_lock list will
          look like t3->t1'->t2'->t1->t2 (where t1' and t2' are TABLE_LIST
          objects added by open_tables() call). There is no flag(or logic) in
          open_tables() that can skip adding these base tables to the list.
          So the logic here should take care of skipping them.

          tables_to_lock_count logic will take care of skipping base tables
          that are added at the end of the list.
          For eg: If RPL_TABLE_LIST is t1->t2->t3, open_tables will modify
          the list into t1->t2->t3->t1'->t2'. t1' and t2' will be skipped
          because tables_to_lock_count logic in this for loop.
        */
        if (table_list_ptr->parent_l)
          continue;
        /*
          We can use a down cast here since we know that every table added
          to the tables_to_lock is a RPL_TABLE_LIST (or child table which is
          skipped above).
        */
        RPL_TABLE_LIST *ptr= static_cast<RPL_TABLE_LIST*>(table_list_ptr);
        DBUG_ASSERT(ptr->m_tabledef_valid);
        TABLE *conv_table;
        if (!ptr->m_tabledef.compatible_with(thd, rgi, ptr->table, &conv_table))
        {
          DBUG_PRINT("debug", ("Table: %s.%s is not compatible with master",
                               ptr->table->s->db.str,
                               ptr->table->s->table_name.str));
          /*
            We should not honour --slave-skip-errors at this point as we are
            having severe errors which should not be skiped.
          */
          thd->is_slave_error= 1;
          /* remove trigger's tables */
          error= ERR_BAD_TABLE_DEF;
          goto err;
        }
        DBUG_PRINT("debug", ("Table: %s.%s is compatible with master"
                             " - conv_table: %p",
                             ptr->table->s->db.str,
                             ptr->table->s->table_name.str, conv_table));
        ptr->m_conv_table= conv_table;
      }
    }

    /*
      ... and then we add all the tables to the table map and but keep
      them in the tables to lock list.

      We also invalidate the query cache for all the tables, since
      they will now be changed.

      TODO [/Matz]: Maybe the query cache should not be invalidated
      here? It might be that a table is not changed, even though it
      was locked for the statement.  We do know that each
      Rows_log_event contain at least one row, so after processing one
      Rows_log_event, we can invalidate the query cache for the
      associated table.
     */
    TABLE_LIST *ptr= rgi->tables_to_lock;
    for (uint i=0 ;  ptr && (i < rgi->tables_to_lock_count); ptr= ptr->next_global, i++)
    {
      /*
        Please see comment in above 'for' loop to know the reason
        for this if condition
      */
      if (ptr->parent_l)
        continue;
      rgi->m_table_map.set_table(ptr->table_id, ptr->table);
      /*
        Following is passing flag about triggers on the server. The problem was
        to pass it between table map event and row event. I do it via extended
        TABLE_LIST (RPL_TABLE_LIST) but row event uses only TABLE so I need to
        find somehow the corresponding TABLE_LIST.
      */
      if (m_table_id == ptr->table_id)
      {
        ptr->table->master_had_triggers=
          ((RPL_TABLE_LIST*)ptr)->master_had_triggers;
      }
    }

5726
#ifdef HAVE_QUERY_CACHE
5727 5728 5729 5730
    /*
      Moved invalidation right before the call to rows_event_stmt_cleanup(),
      to avoid query cache being polluted with stale entries,
    */
5731
# ifdef WITH_WSREP
5732 5733
    /* Query cache is not invalidated on wsrep applier here */
    if (!(WSREP(thd) && wsrep_thd_is_applying(thd)))
5734
# endif /* WITH_WSREP */
Marko Mäkelä's avatar
Marko Mäkelä committed
5735
      query_cache.invalidate_locked_for_write(thd, rgi->tables_to_lock);
5736
#endif /* HAVE_QUERY_CACHE */
5737 5738 5739 5740 5741 5742 5743 5744 5745 5746
  }

  table= m_table= rgi->m_table_map.get_table(m_table_id);

  DBUG_PRINT("debug", ("m_table:%p, m_table_id: %llu%s",
                       m_table, m_table_id,
                       table && master_had_triggers ?
                       " (master had triggers)" : ""));
  if (table)
  {
5747
    Rows_log_event::Db_restore_ctx restore_ctx(this);
5748
    master_had_triggers= table->master_had_triggers;
5749
    bool transactional_table= table->file->has_transactions_and_rollback();
Sachin's avatar
Sachin committed
5750
    table->file->prepare_for_insert(get_general_type_code() != WRITE_ROWS_EVENT);
5751

5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845
    /*
      table == NULL means that this table should not be replicated
      (this was set up by Table_map_log_event::do_apply_event()
      which tested replicate-* rules).
    */

    /*
      It's not needed to set_time() but
      1) it continues the property that "Time" in SHOW PROCESSLIST shows how
      much slave is behind
      2) it will be needed when we allow replication from a table with no
      TIMESTAMP column to a table with one.
      So we call set_time(), like in SBR. Presently it changes nothing.
    */
    thd->set_time(when, when_sec_part);

     if (m_width == table->s->fields && bitmap_is_set_all(&m_cols))
      set_flags(COMPLETE_ROWS_F);

    /* 
      Set tables write and read sets.
      
      Read_set contains all slave columns (in case we are going to fetch
      a complete record from slave)
      
      Write_set equals the m_cols bitmap sent from master but it can be 
      longer if slave has extra columns. 
     */ 

    DBUG_PRINT_BITSET("debug", "Setting table's read_set from: %s", &m_cols);
    
    bitmap_set_all(table->read_set);
    if (get_general_type_code() == DELETE_ROWS_EVENT ||
        get_general_type_code() == UPDATE_ROWS_EVENT)
      bitmap_intersect(table->read_set,&m_cols);

    bitmap_set_all(table->write_set);
    table->rpl_write_set= table->write_set;

    /* WRITE ROWS EVENTS store the bitmap in m_cols instead of m_cols_ai */
    MY_BITMAP *after_image= ((get_general_type_code() == UPDATE_ROWS_EVENT) ?
                             &m_cols_ai : &m_cols);
    bitmap_intersect(table->write_set, after_image);

    this->slave_exec_mode= slave_exec_mode_options; // fix the mode

    // Do event specific preparations 
    error= do_before_row_operations(rli);

    /*
      Bug#56662 Assertion failed: next_insert_id == 0, file handler.cc
      Don't allow generation of auto_increment value when processing
      rows event by setting 'MODE_NO_AUTO_VALUE_ON_ZERO'. The exception
      to this rule happens when the auto_inc column exists on some
      extra columns on the slave. In that case, do not force
      MODE_NO_AUTO_VALUE_ON_ZERO.
    */
    sql_mode_t saved_sql_mode= thd->variables.sql_mode;
    if (!is_auto_inc_in_extra_columns())
      thd->variables.sql_mode= MODE_NO_AUTO_VALUE_ON_ZERO;

    // row processing loop

    /* 
      set the initial time of this ROWS statement if it was not done
      before in some other ROWS event. 
     */
    rgi->set_row_stmt_start_timestamp();

    THD_STAGE_INFO(thd, stage_executing);
    do
    {
      /* in_use can have been set to NULL in close_tables_for_reopen */
      THD* old_thd= table->in_use;
      if (!table->in_use)
        table->in_use= thd;

      error= do_exec_row(rgi);

      if (unlikely(error))
        DBUG_PRINT("info", ("error: %s", HA_ERR(error)));
      DBUG_ASSERT(error != HA_ERR_RECORD_DELETED);

      table->in_use = old_thd;

      if (unlikely(error))
      {
        int actual_error= convert_handler_error(error, thd, table);
        bool idempotent_error= (idempotent_error_code(error) &&
                               (slave_exec_mode == SLAVE_EXEC_MODE_IDEMPOTENT));
        bool ignored_error= (idempotent_error == 0 ?
                             ignored_error_code(actual_error) : 0);

#ifdef WITH_WSREP
5846 5847
        if (WSREP(thd) && thd->wsrep_applier &&
            wsrep_ignored_error_code(this, actual_error))
5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881
        {
          idempotent_error= true;
          thd->wsrep_has_ignored_error= true;
        }
#endif /* WITH_WSREP */
        if (idempotent_error || ignored_error)
        {
          if (global_system_variables.log_warnings)
            slave_rows_error_report(WARNING_LEVEL, error, rgi, thd, table,
                                    get_type_str(),
                                    RPL_LOG_NAME, log_pos);
          thd->clear_error(1);
          error= 0;
          if (idempotent_error == 0)
            break;
        }
      }

      /*
       If m_curr_row_end  was not set during event execution (e.g., because
       of errors) we can't proceed to the next row. If the error is transient
       (i.e., error==0 at this point) we must call unpack_current_row() to set 
       m_curr_row_end.
      */ 
   
      DBUG_PRINT("info", ("curr_row: %p; curr_row_end: %p; rows_end:%p",
                          m_curr_row, m_curr_row_end, m_rows_end));

      if (!m_curr_row_end && likely(!error))
        error= unpack_current_row(rgi);

      m_curr_row= m_curr_row_end;
 
      if (likely(error == 0) && !transactional_table)
5882 5883
        thd->transaction->all.modified_non_trans_table=
          thd->transaction->stmt.modified_non_trans_table= TRUE;
5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898
    } // row processing loop
    while (error == 0 && (m_curr_row != m_rows_end));

    /*
      Restore the sql_mode after the rows event is processed.
    */
    thd->variables.sql_mode= saved_sql_mode;

    {/**
         The following failure injecion works in cooperation with tests 
         setting @@global.debug= 'd,stop_slave_middle_group'.
         The sql thread receives the killed status and will proceed 
         to shutdown trying to finish incomplete events group.
     */
      DBUG_EXECUTE_IF("stop_slave_middle_group",
5899
                      if (thd->transaction->all.modified_non_trans_table)
5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934 5935
                        const_cast<Relay_log_info*>(rli)->abort_slave= 1;);
    }

    if (unlikely(error= do_after_row_operations(rli, error)) &&
        ignored_error_code(convert_handler_error(error, thd, table)))
    {

      if (global_system_variables.log_warnings)
        slave_rows_error_report(WARNING_LEVEL, error, rgi, thd, table,
                                get_type_str(),
                                RPL_LOG_NAME, log_pos);
      thd->clear_error(1);
      error= 0;
    }
  } // if (table)

  
  if (unlikely(error))
  {
    slave_rows_error_report(ERROR_LEVEL, error, rgi, thd, table,
                             get_type_str(),
                             RPL_LOG_NAME, log_pos);
    /*
      @todo We should probably not call
      reset_current_stmt_binlog_format_row() from here.

      Note: this applies to log_event_old.cc too.
      /Sven
    */
    thd->reset_current_stmt_binlog_format_row();
    thd->is_slave_error= 1;
    /* remove trigger's tables */
    goto err;
  }

  /* remove trigger's tables */
Marko Mäkelä's avatar
Marko Mäkelä committed
5936
  restore_empty_query_table_list(thd->lex);
5937 5938

#if defined(WITH_WSREP) && defined(HAVE_QUERY_CACHE)
Marko Mäkelä's avatar
Marko Mäkelä committed
5939 5940
  if (WSREP(thd) && wsrep_thd_is_applying(thd))
    query_cache.invalidate_locked_for_write(thd, rgi->tables_to_lock);
5941 5942
#endif /* WITH_WSREP && HAVE_QUERY_CACHE */

Marko Mäkelä's avatar
Marko Mäkelä committed
5943 5944 5945 5946 5947 5948 5949 5950 5951 5952
  if (get_flags(STMT_END_F))
  {
    if (unlikely((error= rows_event_stmt_cleanup(rgi, thd))))
      slave_rows_error_report(ERROR_LEVEL, thd->is_error() ? 0 : error,
                              rgi, thd, table, get_type_str(),
                              RPL_LOG_NAME, log_pos);
    if (thd->slave_thread)
      free_root(thd->mem_root, MYF(MY_KEEP_PREALLOC));
  }

5953 5954 5955
  DBUG_RETURN(error);

err:
Marko Mäkelä's avatar
Marko Mäkelä committed
5956
  restore_empty_query_table_list(thd->lex);
5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016
  rgi->slave_close_thread_tables(thd);
  DBUG_RETURN(error);
}

Log_event::enum_skip_reason
Rows_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    If the slave skip counter is 1 and this event does not end a
    statement, then we should not start executing on the next event.
    Otherwise, we defer the decision to the normal skipping logic.
  */
  if (rgi->rli->slave_skip_counter == 1 && !get_flags(STMT_END_F))
    return Log_event::EVENT_SKIP_IGNORE;
  else
    return Log_event::do_shall_skip(rgi);
}

/**
   The function is called at Rows_log_event statement commit time,
   normally from Rows_log_event::do_update_pos() and possibly from
   Query_log_event::do_apply_event() of the COMMIT.
   The function commits the last statement for engines, binlog and
   releases resources have been allocated for the statement.
  
   @retval  0         Ok.
   @retval  non-zero  Error at the commit.
 */

static int rows_event_stmt_cleanup(rpl_group_info *rgi, THD * thd)
{
  int error;
  DBUG_ENTER("rows_event_stmt_cleanup");

  {
    /*
      This is the end of a statement or transaction, so close (and
      unlock) the tables we opened when processing the
      Table_map_log_event starting the statement.

      OBSERVER.  This will clear *all* mappings, not only those that
      are open for the table. There is not good handle for on-close
      actions for tables.

      NOTE. Even if we have no table ('table' == 0) we still need to be
      here, so that we increase the group relay log position. If we didn't, we
      could have a group relay log position which lags behind "forever"
      (assume the last master's transaction is ignored by the slave because of
      replicate-ignore rules).
    */
    error= thd->binlog_flush_pending_rows_event(TRUE);

    /*
      If this event is not in a transaction, the call below will, if some
      transactional storage engines are involved, commit the statement into
      them and flush the pending event to binlog.
      If this event is in a transaction, the call will do nothing, but a
      Xid_log_event will come next which will, if some transactional engines
      are involved, commit the transaction and flush the pending event to the
      binlog.
6017 6018 6019
      We check for thd->transaction_rollback_request because it is possible
      there was a deadlock that was ignored by slave-skip-errors. Normally, the
      deadlock would have been rolled back already.
6020
    */
6021 6022 6023
    error|= (int) ((error || thd->transaction_rollback_request)
                       ? trans_rollback_stmt(thd)
                       : trans_commit_stmt(thd));
6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052

    /*
      Now what if this is not a transactional engine? we still need to
      flush the pending event to the binlog; we did it with
      thd->binlog_flush_pending_rows_event(). Note that we imitate
      what is done for real queries: a call to
      ha_autocommit_or_rollback() (sometimes only if involves a
      transactional engine), and a call to be sure to have the pending
      event flushed.
    */

    /*
      @todo We should probably not call
      reset_current_stmt_binlog_format_row() from here.

      Note: this applies to log_event_old.cc too

      Btw, the previous comment about transactional engines does not
      seem related to anything that happens here.
      /Sven
    */
    thd->reset_current_stmt_binlog_format_row();

    /*
      Reset modified_non_trans_table that we have set in
      rows_log_event::do_apply_event()
    */
    if (!thd->in_multi_stmt_transaction_mode())
    {
6053 6054
      thd->transaction->all.modified_non_trans_table= 0;
      thd->transaction->all.m_unsafe_rollback_flags&= ~THD_TRANS::DID_WAIT;
6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111
    }

    rgi->cleanup_context(thd, 0);
  }
  DBUG_RETURN(error);
}

/**
   The method either increments the relay log position or
   commits the current statement and increments the master group 
   possition if the event is STMT_END_F flagged and
   the statement corresponds to the autocommit query (i.e replicated
   without wrapping in BEGIN/COMMIT)

   @retval 0         Success
   @retval non-zero  Error in the statement commit
 */
int
Rows_log_event::do_update_pos(rpl_group_info *rgi)
{
  Relay_log_info *rli= rgi->rli;
  int error= 0;
  DBUG_ENTER("Rows_log_event::do_update_pos");

  DBUG_PRINT("info", ("flags: %s",
                      get_flags(STMT_END_F) ? "STMT_END_F " : ""));

  if (get_flags(STMT_END_F))
  {
    /*
      Indicate that a statement is finished.
      Step the group log position if we are not in a transaction,
      otherwise increase the event log position.
    */
    error= rli->stmt_done(log_pos, thd, rgi);
    /*
      Clear any errors in thd->net.last_err*. It is not known if this is
      needed or not. It is believed that any errors that may exist in
      thd->net.last_err* are allowed. Examples of errors are "key not
      found", which is produced in the test case rpl_row_conflicts.test
    */
    thd->clear_error();
  }
  else
  {
    rgi->inc_event_relay_log_pos();
  }

  DBUG_RETURN(error);
}

#endif /* defined(HAVE_REPLICATION) */


bool Rows_log_event::write_data_header()
{
  uchar buf[ROWS_HEADER_LEN_V2];        // No need to init the buffer
6112
  DBUG_ASSERT(m_table_id != UINT32_MAX);
6113 6114
  DBUG_EXECUTE_IF("old_row_based_repl_4_byte_map_id_master",
                  {
6115
                    int4store(buf + 0, (ulong) m_table_id);
6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133
                    int2store(buf + 4, m_flags);
                    return (write_data(buf, 6));
                  });
  int6store(buf + RW_MAPID_OFFSET, m_table_id);
  int2store(buf + RW_FLAGS_OFFSET, m_flags);
  return write_data(buf, ROWS_HEADER_LEN);
}

bool Rows_log_event::write_data_body()
{
  /*
     Note that this should be the number of *bits*, not the number of
     bytes.
  */
  uchar sbuf[MAX_INT_WIDTH];
  my_ptrdiff_t const data_size= m_rows_cur - m_rows_buf;
  bool res= false;
  uchar *const sbuf_end= net_store_length(sbuf, (size_t) m_width);
6134 6135
  uint bitmap_size= no_bytes_in_export_map(&m_cols);
  uchar *bitmap;
6136 6137 6138 6139 6140
  DBUG_ASSERT(static_cast<size_t>(sbuf_end - sbuf) <= sizeof(sbuf));

  DBUG_DUMP("m_width", sbuf, (size_t) (sbuf_end - sbuf));
  res= res || write_data(sbuf, (size_t) (sbuf_end - sbuf));

6141 6142 6143 6144 6145
  bitmap= (uchar*) my_alloca(bitmap_size);
  bitmap_export(bitmap, &m_cols);

  DBUG_DUMP("m_cols", bitmap, bitmap_size);
  res= res || write_data(bitmap, bitmap_size);
6146 6147 6148 6149 6150
  /*
    TODO[refactor write]: Remove the "down cast" here (and elsewhere).
   */
  if (get_general_type_code() == UPDATE_ROWS_EVENT)
  {
6151 6152 6153 6154 6155
    DBUG_ASSERT(m_cols.n_bits == m_cols_ai.n_bits);
    bitmap_export(bitmap, &m_cols_ai);

    DBUG_DUMP("m_cols_ai", bitmap, bitmap_size);
    res= res || write_data(bitmap, bitmap_size);
6156 6157 6158
  }
  DBUG_DUMP("rows", m_rows_buf, data_size);
  res= res || write_data(m_rows_buf, (size_t) data_size);
6159
  my_afree(bitmap);
6160 6161 6162 6163

  return res;
}

6164

6165 6166
bool Rows_log_event::write_compressed()
{
6167 6168 6169
  uchar *m_rows_buf_tmp= m_rows_buf;
  uchar *m_rows_cur_tmp= m_rows_cur;
  bool ret= true;
6170
  uint32 comlen, alloc_size;
6171 6172 6173
  comlen= alloc_size= binlog_get_compress_len((uint32)(m_rows_cur_tmp -
                                                       m_rows_buf_tmp));
  m_rows_buf= (uchar*) my_safe_alloca(alloc_size);
6174
  if(m_rows_buf &&
6175
     !binlog_buf_compress(m_rows_buf_tmp, m_rows_buf,
6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304
                          (uint32)(m_rows_cur_tmp - m_rows_buf_tmp), &comlen))
  {
    m_rows_cur= comlen + m_rows_buf;
    ret= Log_event::write();
  }
  my_safe_afree(m_rows_buf, alloc_size);
  m_rows_buf= m_rows_buf_tmp;
  m_rows_cur= m_rows_cur_tmp;
  return ret;
}


#if defined(HAVE_REPLICATION)
void Rows_log_event::pack_info(Protocol *protocol)
{
  char buf[256];
  char const *const flagstr=
    get_flags(STMT_END_F) ? " flags: STMT_END_F" : "";
  size_t bytes= my_snprintf(buf, sizeof(buf),
                               "table_id: %llu%s", m_table_id, flagstr);
  protocol->store(buf, bytes, &my_charset_bin);
}
#endif


/**************************************************************************
	Annotate_rows_log_event member functions
**************************************************************************/

Annotate_rows_log_event::Annotate_rows_log_event(THD *thd,
                                                 bool using_trans,
                                                 bool direct)
  : Log_event(thd, 0, using_trans),
    m_save_thd_query_txt(0),
    m_save_thd_query_len(0),
    m_saved_thd_query(false),
    m_used_query_txt(0)
{
  m_query_txt= thd->query();
  m_query_len= thd->query_length();
  if (direct)
    cache_type= Log_event::EVENT_NO_CACHE;
}


bool Annotate_rows_log_event::write_data_header()
{ 
  return 0;
}


bool Annotate_rows_log_event::write_data_body()
{
  return write_data(m_query_txt, m_query_len);
}


#if defined(HAVE_REPLICATION)
void Annotate_rows_log_event::pack_info(Protocol* protocol)
{
  if (m_query_txt && m_query_len)
    protocol->store(m_query_txt, m_query_len, &my_charset_bin);
}
#endif


#if defined(HAVE_REPLICATION)
int Annotate_rows_log_event::do_apply_event(rpl_group_info *rgi)
{
  rgi->free_annotate_event();
  m_save_thd_query_txt= thd->query();
  m_save_thd_query_len= thd->query_length();
  m_saved_thd_query= true;
  m_used_query_txt= 1;
  thd->set_query(m_query_txt, m_query_len);
  return 0;
}
#endif


#if defined(HAVE_REPLICATION)
int Annotate_rows_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}
#endif


#if defined(HAVE_REPLICATION)
Log_event::enum_skip_reason
Annotate_rows_log_event::do_shall_skip(rpl_group_info *rgi)
{
  return continue_group(rgi);
}
#endif

/**************************************************************************
	Table_map_log_event member functions and support functions
**************************************************************************/

/**
  Save the field metadata based on the real_type of the field.
  The metadata saved depends on the type of the field. Some fields
  store a single byte for pack_length() while others store two bytes
  for field_length (max length).
  
  @retval  0  Ok.

  @todo
  We may want to consider changing the encoding of the information.
  Currently, the code attempts to minimize the number of bytes written to 
  the tablemap. There are at least two other alternatives; 1) using 
  net_store_length() to store the data allowing it to choose the number of
  bytes that are appropriate thereby making the code much easier to 
  maintain (only 1 place to change the encoding), or 2) use a fixed number
  of bytes for each field. The problem with option 1 is that net_store_length()
  will use one byte if the value < 251, but 3 bytes if it is > 250. Thus,
  for fields like CHAR which can be no larger than 255 characters, the method
  will use 3 bytes when the value is > 250. Further, every value that is
  encoded using 2 parts (e.g., pack_length, field_length) will be numerically
  > 250 therefore will use 3 bytes for eah value. The problem with option 2
  is less wasteful for space but does waste 1 byte for every field that does
  not encode 2 parts. 
*/
int Table_map_log_event::save_field_metadata()
{
  DBUG_ENTER("Table_map_log_event::save_field_metadata");
  int index= 0;
6305
  Binlog_type_info *info;
6306 6307 6308
  for (unsigned int i= 0 ; i < m_table->s->fields ; i++)
  {
    DBUG_PRINT("debug", ("field_type: %d", m_coltype[i]));
6309
    info= binlog_type_info_array + i;
6310
    int2store(&m_field_metadata[index], info->m_metadata);
6311 6312 6313 6314 6315 6316
    index+= info->m_metadata_size;
    DBUG_EXECUTE_IF("inject_invalid_blob_size",
                    {
                      if (m_coltype[i] == MYSQL_TYPE_BLOB)
                        m_field_metadata[index-1] = 5;
                    });
6317 6318 6319 6320 6321 6322 6323 6324 6325 6326
  }
  DBUG_RETURN(index);
}


/*
  Constructor used to build an event for writing to the binary log.
  Mats says tbl->s lives longer than this event so it's ok to copy pointers
  (tbl->s->db etc) and not pointer content.
 */
6327
Table_map_log_event::Table_map_log_event(THD *thd, TABLE *tbl, ulonglong tid,
6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342
                                         bool is_transactional)
  : Log_event(thd, 0, is_transactional),
    m_table(tbl),
    m_dbnam(tbl->s->db.str),
    m_dblen(m_dbnam ? tbl->s->db.length : 0),
    m_tblnam(tbl->s->table_name.str),
    m_tbllen(tbl->s->table_name.length),
    m_colcnt(tbl->s->fields),
    m_memory(NULL),
    m_table_id(tid),
    m_flags(TM_BIT_LEN_EXACT_F),
    m_data_size(0),
    m_field_metadata(0),
    m_field_metadata_size(0),
    m_null_bits(0),
6343 6344 6345
    m_meta_memory(NULL),
    m_optional_metadata_len(0),
    m_optional_metadata(NULL)
6346 6347 6348 6349
{
  uchar cbuf[MAX_INT_WIDTH];
  uchar *cbuf_end;
  DBUG_ENTER("Table_map_log_event::Table_map_log_event(TABLE)");
6350
  DBUG_ASSERT(m_table_id != UINT32_MAX);
6351 6352 6353 6354 6355 6356 6357 6358 6359 6360
  /*
    In TABLE_SHARE, "db" and "table_name" are 0-terminated (see this comment in
    table.cc / alloc_table_share():
      Use the fact the key is db/0/table_name/0
    As we rely on this let's assert it.
  */
  DBUG_ASSERT((tbl->s->db.str == 0) ||
              (tbl->s->db.str[tbl->s->db.length] == 0));
  DBUG_ASSERT(tbl->s->table_name.str[tbl->s->table_name.length] == 0);

6361 6362 6363 6364
  binlog_type_info_array= (Binlog_type_info *)thd->alloc(m_table->s->fields *
                                                   sizeof(Binlog_type_info));
  for (uint i= 0; i <  m_table->s->fields; i++)
    binlog_type_info_array[i]= m_table->field[i]->binlog_type_info();
6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377

  m_data_size=  TABLE_MAP_HEADER_LEN;
  DBUG_EXECUTE_IF("old_row_based_repl_4_byte_map_id_master", m_data_size= 6;);
  m_data_size+= m_dblen + 2;	// Include length and terminating \0
  m_data_size+= m_tbllen + 2;	// Include length and terminating \0
  cbuf_end= net_store_length(cbuf, (size_t) m_colcnt);
  DBUG_ASSERT(static_cast<size_t>(cbuf_end - cbuf) <= sizeof(cbuf));
  m_data_size+= (cbuf_end - cbuf) + m_colcnt;	// COLCNT and column types

  if (tbl->triggers)
    m_flags|= TM_BIT_HAS_TRIGGERS_F;

  /* If malloc fails, caught in is_valid() */
6378
  if ((m_memory= (uchar*) my_malloc(PSI_INSTRUMENT_ME, m_colcnt, MYF(MY_WME))))
6379 6380 6381
  {
    m_coltype= reinterpret_cast<uchar*>(m_memory);
    for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
6382 6383
      m_coltype[i]= binlog_type_info_array[i].m_type_code;
    DBUG_EXECUTE_IF("inject_invalid_column_type", m_coltype[1]= 230;);
6384 6385 6386 6387 6388 6389 6390 6391 6392 6393
  }

  /*
    Calculate a bitmap for the results of maybe_null() for all columns.
    The bitmap is used to determine when there is a column from the master
    that is not on the slave and is null and thus not in the row data during
    replication.
  */
  uint num_null_bytes= (m_table->s->fields + 7) / 8;
  m_data_size+= num_null_bytes;
6394
  m_meta_memory= (uchar *)my_multi_malloc(PSI_INSTRUMENT_ME, MYF(MY_WME),
6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420 6421
                                 &m_null_bits, num_null_bytes,
                                 &m_field_metadata, (m_colcnt * 2),
                                 NULL);

  bzero(m_field_metadata, (m_colcnt * 2));

  /*
    Create an array for the field metadata and store it.
  */
  m_field_metadata_size= save_field_metadata();
  DBUG_ASSERT(m_field_metadata_size <= (m_colcnt * 2));

  /*
    Now set the size of the data to the size of the field metadata array
    plus one or three bytes (see pack.c:net_store_length) for number of 
    elements in the field metadata array.
  */
  if (m_field_metadata_size < 251)
    m_data_size+= m_field_metadata_size + 1; 
  else
    m_data_size+= m_field_metadata_size + 3; 

  bzero(m_null_bits, num_null_bytes);
  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
    if (m_table->field[i]->maybe_null())
      m_null_bits[(i / 8)]+= 1 << (i % 8);

6422 6423 6424
  init_metadata_fields();
  m_data_size+= m_metadata_buf.length();

6425 6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544 6545
  DBUG_VOID_RETURN;
}


/*
  Return value is an error code, one of:

      -1     Failure to open table   [from open_tables()]
       0     Success
       1     No room for more tables [from set_table()]
       2     Out of memory           [from set_table()]
       3     Wrong table definition
       4     Daisy-chaining RBR with SBR not possible
 */

#if defined(HAVE_REPLICATION)

enum enum_tbl_map_status
{
  /* no duplicate identifier found */
  OK_TO_PROCESS= 0,

  /* this table map must be filtered out */
  FILTERED_OUT= 1,

  /* identifier mapping table with different properties */
  SAME_ID_MAPPING_DIFFERENT_TABLE= 2,
  
  /* a duplicate identifier was found mapping the same table */
  SAME_ID_MAPPING_SAME_TABLE= 3
};

/*
  Checks if this table map event should be processed or not. First
  it checks the filtering rules, and then looks for duplicate identifiers
  in the existing list of rli->tables_to_lock.

  It checks that there hasn't been any corruption by verifying that there
  are no duplicate entries with different properties.

  In some cases, some binary logs could get corrupted, showing several
  tables mapped to the same table_id, 0 (see: BUG#56226). Thus we do this
  early sanity check for such cases and avoid that the server crashes 
  later.

  In some corner cases, the master logs duplicate table map events, i.e.,
  same id, same database name, same table name (see: BUG#37137). This is
  different from the above as it's the same table that is mapped again 
  to the same identifier. Thus we cannot just check for same ids and 
  assume that the event is corrupted we need to check every property. 

  NOTE: in the event that BUG#37137 ever gets fixed, this extra check 
        will still be valid because we would need to support old binary 
        logs anyway.

  @param rli The relay log info reference.
  @param table_list A list element containing the table to check against.
  @return OK_TO_PROCESS 
            if there was no identifier already in rli->tables_to_lock 
            
          FILTERED_OUT
            if the event is filtered according to the filtering rules

          SAME_ID_MAPPING_DIFFERENT_TABLE 
            if the same identifier already maps a different table in 
            rli->tables_to_lock

          SAME_ID_MAPPING_SAME_TABLE 
            if the same identifier already maps the same table in 
            rli->tables_to_lock.
*/
static enum_tbl_map_status
check_table_map(rpl_group_info *rgi, RPL_TABLE_LIST *table_list)
{
  DBUG_ENTER("check_table_map");
  enum_tbl_map_status res= OK_TO_PROCESS;
  Relay_log_info *rli= rgi->rli;
  if ((rgi->thd->slave_thread /* filtering is for slave only */ ||
        IF_WSREP((WSREP(rgi->thd) && rgi->thd->wsrep_applier), 0)) &&
      (!rli->mi->rpl_filter->db_ok(table_list->db.str) ||
       (rli->mi->rpl_filter->is_on() && !rli->mi->rpl_filter->tables_ok("", table_list))))
    res= FILTERED_OUT;
  else
  {
    RPL_TABLE_LIST *ptr= static_cast<RPL_TABLE_LIST*>(rgi->tables_to_lock);
    for(uint i=0 ; ptr && (i< rgi->tables_to_lock_count); 
        ptr= static_cast<RPL_TABLE_LIST*>(ptr->next_local), i++)
    {
      if (ptr->table_id == table_list->table_id)
      {

        if (cmp(&ptr->db, &table_list->db) ||
            cmp(&ptr->alias, &table_list->table_name) ||
            ptr->lock_type != TL_WRITE) // the ::do_apply_event always sets TL_WRITE
          res= SAME_ID_MAPPING_DIFFERENT_TABLE;
        else
          res= SAME_ID_MAPPING_SAME_TABLE;

        break;
      }
    }
  }

  DBUG_PRINT("debug", ("check of table map ended up with: %u", res));

  DBUG_RETURN(res);
}

int Table_map_log_event::do_apply_event(rpl_group_info *rgi)
{
  RPL_TABLE_LIST *table_list;
  char *db_mem, *tname_mem, *ptr;
  size_t dummy_len, db_mem_length, tname_mem_length;
  void *memory;
  Rpl_filter *filter;
  Relay_log_info const *rli= rgi->rli;
  DBUG_ENTER("Table_map_log_event::do_apply_event(Relay_log_info*)");

  /* Step the query id to mark what columns that are actually used. */
  thd->set_query_id(next_query_id());

6546
  if (!(memory= my_multi_malloc(PSI_INSTRUMENT_ME, MYF(MY_WME),
6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635
                                &table_list, (uint) sizeof(RPL_TABLE_LIST),
                                &db_mem, (uint) NAME_LEN + 1,
                                &tname_mem, (uint) NAME_LEN + 1,
                                NullS)))
    DBUG_RETURN(HA_ERR_OUT_OF_MEM);

  db_mem_length= strmov(db_mem, m_dbnam) - db_mem;
  tname_mem_length= strmov(tname_mem, m_tblnam) - tname_mem;
  if (lower_case_table_names)
  {
    my_casedn_str(files_charset_info, (char*)tname_mem);
    my_casedn_str(files_charset_info, (char*)db_mem);
  }

  /* call from mysql_client_binlog_statement() will not set rli->mi */
  filter= rgi->thd->slave_thread ? rli->mi->rpl_filter : global_rpl_filter;

  /* rewrite rules changed the database */
  if (((ptr= (char*) filter->get_rewrite_db(db_mem, &dummy_len)) != db_mem))
    db_mem_length= strmov(db_mem, ptr) - db_mem;

  LEX_CSTRING tmp_db_name=  {db_mem, db_mem_length };
  LEX_CSTRING tmp_tbl_name= {tname_mem, tname_mem_length };

  table_list->init_one_table(&tmp_db_name, &tmp_tbl_name, 0, TL_WRITE);
  table_list->table_id= DBUG_EVALUATE_IF("inject_tblmap_same_id_maps_diff_table", 0, m_table_id);
  table_list->updating= 1;
  table_list->required_type= TABLE_TYPE_NORMAL;

  DBUG_PRINT("debug", ("table: %s is mapped to %llu",
                       table_list->table_name.str,
                       table_list->table_id));
  table_list->master_had_triggers= ((m_flags & TM_BIT_HAS_TRIGGERS_F) ? 1 : 0);
  DBUG_PRINT("debug", ("table->master_had_triggers=%d", 
                       (int)table_list->master_had_triggers));

  enum_tbl_map_status tblmap_status= check_table_map(rgi, table_list);
  if (tblmap_status == OK_TO_PROCESS)
  {
    DBUG_ASSERT(thd->lex->query_tables != table_list);

    /*
      Use placement new to construct the table_def instance in the
      memory allocated for it inside table_list.

      The memory allocated by the table_def structure (i.e., not the
      memory allocated *for* the table_def structure) is released
      inside Relay_log_info::clear_tables_to_lock() by calling the
      table_def destructor explicitly.
    */
    new (&table_list->m_tabledef)
      table_def(m_coltype, m_colcnt,
                m_field_metadata, m_field_metadata_size,
                m_null_bits, m_flags);
    table_list->m_tabledef_valid= TRUE;
    table_list->m_conv_table= NULL;
    table_list->open_type= OT_BASE_ONLY;

    /*
      We record in the slave's information that the table should be
      locked by linking the table into the list of tables to lock.
    */
    table_list->next_global= table_list->next_local= rgi->tables_to_lock;
    rgi->tables_to_lock= table_list;
    rgi->tables_to_lock_count++;
    /* 'memory' is freed in clear_tables_to_lock */
  }
  else  // FILTERED_OUT, SAME_ID_MAPPING_*
  {
    /*
      If mapped already but with different properties, we raise an
      error.
      If mapped already but with same properties we skip the event.
      If filtered out we skip the event.

      In all three cases, we need to free the memory previously 
      allocated.
     */
    if (tblmap_status == SAME_ID_MAPPING_DIFFERENT_TABLE)
    {
      /*
        Something bad has happened. We need to stop the slave as strange things
        could happen if we proceed: slave crash, wrong table being updated, ...
        As a consequence we push an error in this case.
       */

      char buf[256];

      my_snprintf(buf, sizeof(buf), 
6636
                  "Found table map event mapping table id %llu which "
6637 6638 6639 6640 6641 6642 6643 6644 6645 6646 6647 6648 6649 6650 6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676
                  "was already mapped but with different settings.",
                  table_list->table_id);

      if (thd->slave_thread)
        rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR, rgi->gtid_info(),
                    ER_THD(thd, ER_SLAVE_FATAL_ERROR), buf);
      else
        /* 
          For the cases in which a 'BINLOG' statement is set to 
          execute in a user session 
         */
        my_error(ER_SLAVE_FATAL_ERROR, MYF(0), buf);
    } 
    
    my_free(memory);
  }

  DBUG_RETURN(tblmap_status == SAME_ID_MAPPING_DIFFERENT_TABLE);
}

Log_event::enum_skip_reason
Table_map_log_event::do_shall_skip(rpl_group_info *rgi)
{
  /*
    If the slave skip counter is 1, then we should not start executing
    on the next event.
  */
  return continue_group(rgi);
}

int Table_map_log_event::do_update_pos(rpl_group_info *rgi)
{
  rgi->inc_event_relay_log_pos();
  return 0;
}

#endif /* defined(HAVE_REPLICATION) */

bool Table_map_log_event::write_data_header()
{
6677
  DBUG_ASSERT(m_table_id != UINT32_MAX);
6678 6679 6680
  uchar buf[TABLE_MAP_HEADER_LEN];
  DBUG_EXECUTE_IF("old_row_based_repl_4_byte_map_id_master",
                  {
6681
                    int4store(buf + 0, (ulong) m_table_id);
6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716 6717 6718
                    int2store(buf + 4, m_flags);
                    return (write_data(buf, 6));
                  });
  int6store(buf + TM_MAPID_OFFSET, m_table_id);
  int2store(buf + TM_FLAGS_OFFSET, m_flags);
  return write_data(buf, TABLE_MAP_HEADER_LEN);
}

bool Table_map_log_event::write_data_body()
{
  DBUG_ASSERT(m_dbnam != NULL);
  DBUG_ASSERT(m_tblnam != NULL);
  /* We use only one byte per length for storage in event: */
  DBUG_ASSERT(m_dblen <= MY_MIN(NAME_LEN, 255));
  DBUG_ASSERT(m_tbllen <= MY_MIN(NAME_LEN, 255));

  uchar const dbuf[]= { (uchar) m_dblen };
  uchar const tbuf[]= { (uchar) m_tbllen };

  uchar cbuf[MAX_INT_WIDTH];
  uchar *const cbuf_end= net_store_length(cbuf, (size_t) m_colcnt);
  DBUG_ASSERT(static_cast<size_t>(cbuf_end - cbuf) <= sizeof(cbuf));

  /*
    Store the size of the field metadata.
  */
  uchar mbuf[MAX_INT_WIDTH];
  uchar *const mbuf_end= net_store_length(mbuf, m_field_metadata_size);

  return write_data(dbuf,      sizeof(dbuf)) ||
         write_data(m_dbnam,   m_dblen+1) ||
         write_data(tbuf,      sizeof(tbuf)) ||
         write_data(m_tblnam,  m_tbllen+1) ||
         write_data(cbuf, (size_t) (cbuf_end - cbuf)) ||
         write_data(m_coltype, m_colcnt) ||
         write_data(mbuf, (size_t) (mbuf_end - mbuf)) ||
         write_data(m_field_metadata, m_field_metadata_size),
6719 6720 6721
         write_data(m_null_bits, (m_colcnt + 7) / 8) ||
         write_data((const uchar*) m_metadata_buf.ptr(),
                                  m_metadata_buf.length());
6722 6723
 }

6724 6725 6726 6727 6728 6729 6730 6731 6732 6733 6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762 6763 6764 6765 6766 6767 6768 6769 6770 6771 6772 6773 6774 6775 6776 6777 6778 6779 6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791 6792 6793 6794 6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833
/**
   stores an integer into packed format.

   @param[out] str_buf  a buffer where the packed integer will be stored.
   @param[in] length  the integer will be packed.
 */
static inline
void store_compressed_length(String &str_buf, ulonglong length)
{
  // Store Type and packed length
  uchar buf[4];
  uchar *buf_ptr = net_store_length(buf, length);

  str_buf.append(reinterpret_cast<char *>(buf), buf_ptr-buf);
}

/**
  Write data into str_buf with Type|Length|Value(TLV) format.

  @param[out] str_buf a buffer where the field is stored.
  @param[in] type  type of the field
  @param[in] length  length of the field value
  @param[in] value  value of the field
*/
static inline
bool write_tlv_field(String &str_buf,
                     enum Table_map_log_event::Optional_metadata_field_type
                     type, uint length, const uchar *value)
{
  /* type is stored in one byte, so it should never bigger than 255. */
  DBUG_ASSERT(static_cast<int>(type) <= 255);
  str_buf.append((char) type);
  store_compressed_length(str_buf, length);
  return str_buf.append(reinterpret_cast<const char *>(value), length);
}

/**
  Write data into str_buf with Type|Length|Value(TLV) format.

  @param[out] str_buf a buffer where the field is stored.
  @param[in] type  type of the field
  @param[in] value  value of the field
*/
static inline
bool write_tlv_field(String &str_buf,
                     enum Table_map_log_event::Optional_metadata_field_type
                     type, const String &value)
{
  return write_tlv_field(str_buf, type, value.length(),
                         reinterpret_cast<const uchar *>(value.ptr()));
}

static inline bool is_character_field(Binlog_type_info *info_array, Field *field)
{
  Binlog_type_info *info= info_array + field->field_index;
  if (!info->m_cs)
    return 0;
  if (info->m_set_typelib || info->m_enum_typelib)
    return 0;
  return 1;
}

static inline bool is_enum_or_set_field(Binlog_type_info *info_array, Field *field) {
  Binlog_type_info *info= info_array + field->field_index;
  if (info->m_set_typelib || info->m_enum_typelib)
    return 1;
  return 0;
}


void Table_map_log_event::init_metadata_fields()
{
  DBUG_ENTER("init_metadata_fields");
  DBUG_EXECUTE_IF("simulate_no_optional_metadata", DBUG_VOID_RETURN;);

  if (binlog_row_metadata == BINLOG_ROW_METADATA_NO_LOG)
    DBUG_VOID_RETURN;
  if (init_signedness_field() ||
      init_charset_field(&is_character_field, DEFAULT_CHARSET,
                         COLUMN_CHARSET) ||
      init_geometry_type_field())
  {
    m_metadata_buf.length(0);
    DBUG_VOID_RETURN;
  }

  if (binlog_row_metadata == BINLOG_ROW_METADATA_FULL)
  {
    if (DBUG_EVALUATE_IF("dont_log_column_name", 0, init_column_name_field()) ||
        init_charset_field(&is_enum_or_set_field, ENUM_AND_SET_DEFAULT_CHARSET,
                           ENUM_AND_SET_COLUMN_CHARSET) ||
        init_set_str_value_field() ||
        init_enum_str_value_field() ||
        init_primary_key_field())
      m_metadata_buf.length(0);
  }
  DBUG_VOID_RETURN;
}

bool Table_map_log_event::init_signedness_field()
{
  /* use it to store signed flags, each numeric column take a bit. */
  StringBuffer<128> buf;
  unsigned char flag= 0;
  unsigned char mask= 0x80;
  Binlog_type_info *info;

  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    info= binlog_type_info_array + i;
6834
    if (info->m_signedness != Binlog_type_info::SIGN_NOT_APPLICABLE)
6835
    {
6836
      if (info->m_signedness == Binlog_type_info::SIGN_UNSIGNED)
6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948
        flag|= mask;
      mask >>= 1;

      // 8 fields are tested, store the result and clear the flag.
      if (mask == 0)
      {
        buf.append(flag);
        flag= 0;
        mask= 0x80;
      }
    }
  }

  // Stores the signedness flags of last few columns
  if (mask != 0x80)
    buf.append(flag);

  // The table has no numeric column, so don't log SIGNEDNESS field
  if (buf.is_empty())
    return false;

  return write_tlv_field(m_metadata_buf, SIGNEDNESS, buf);
}

bool Table_map_log_event::init_charset_field(
    bool (* include_type)(Binlog_type_info *, Field *),
    Optional_metadata_field_type default_charset_type,
    Optional_metadata_field_type column_charset_type)
{
  DBUG_EXECUTE_IF("simulate_init_charset_field_error", return true;);

  std::map<uint, uint> collation_map;
  // For counting characters columns
  uint char_col_cnt= 0;

  /* Find the collation number used by most fields */
  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    if ((*include_type)(binlog_type_info_array, m_table->field[i]))
    {
      collation_map[binlog_type_info_array[i].m_cs->number]++;
      char_col_cnt++;
    }
  }

  if (char_col_cnt == 0)
    return false;

  /* Find the most used collation */
  uint most_used_collation= 0;
  uint most_used_count= 0;
  for (std::map<uint, uint>::iterator it= collation_map.begin();
       it != collation_map.end(); it++)
  {
    if (it->second > most_used_count)
    {
      most_used_count= it->second;
      most_used_collation= it->first;
    }
  }

  /*
    Comparing length of COLUMN_CHARSET field and COLUMN_CHARSET_WITH_DEFAULT
    field to decide which field should be logged.

    Length of COLUMN_CHARSET = character column count * collation id size.
    Length of COLUMN_CHARSET_WITH_DEFAULT =
     default collation_id size + count of columns not use default charset *
     (column index size + collation id size)

    Assume column index just uses 1 byte and collation number also uses 1 byte.
  */
  if (char_col_cnt * 1 < (1 + (char_col_cnt - most_used_count) * 2))
  {
    StringBuffer<512> buf;

    /*
      Stores character set information into COLUMN_CHARSET format,
      character sets of all columns are stored one by one.
      -----------------------------------------
      | Charset number | .... |Charset number |
      -----------------------------------------
    */
    for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
    {
      if (include_type(binlog_type_info_array, m_table->field[i]))
        store_compressed_length(buf, binlog_type_info_array[i].m_cs->number);
    }
    return write_tlv_field(m_metadata_buf, column_charset_type, buf);
  }
  else
  {
    StringBuffer<512> buf;
    uint char_column_index= 0;
    uint default_collation= most_used_collation;

    /*
      Stores character set information into DEFAULT_CHARSET format,
      First stores the default character set, and then stores the character
      sets different to default character with their column index one by one.
      --------------------------------------------------------
      | Default Charset | Col Index | Charset number | ...   |
      --------------------------------------------------------
    */

    // Store the default collation number
    store_compressed_length(buf, default_collation);

    for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
    {
      if (include_type(binlog_type_info_array, m_table->field[i]))
      {
6949 6950 6951
        CHARSET_INFO *cs= binlog_type_info_array[i].m_cs;
        DBUG_ASSERT(cs);
        if (cs->number != default_collation)
6952 6953
        {
          store_compressed_length(buf, char_column_index);
6954
          store_compressed_length(buf, cs->number);
6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970 6971 6972 6973 6974 6975 6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994 6995 6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105
        }
        char_column_index++;
      }
    }
    return write_tlv_field(m_metadata_buf, default_charset_type, buf);
  }
}

bool Table_map_log_event::init_column_name_field()
{
  StringBuffer<2048> buf;

  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    size_t len= m_table->field[i]->field_name.length;

    store_compressed_length(buf, len);
    buf.append(m_table->field[i]->field_name.str, len);
  }
  return write_tlv_field(m_metadata_buf, COLUMN_NAME, buf);
}

bool Table_map_log_event::init_set_str_value_field()
{
  StringBuffer<1024> buf;
  TYPELIB *typelib;

  /*
    SET string values are stored in the same format:
    ----------------------------------------------
    | Value number | value1 len | value 1|  .... |  // first SET column
    ----------------------------------------------
    | Value number | value1 len | value 1|  .... |  // second SET column
    ----------------------------------------------
   */
  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    if ((typelib= binlog_type_info_array[i].m_set_typelib))
    {
      store_compressed_length(buf, typelib->count);
      for (unsigned int i= 0; i < typelib->count; i++)
      {
        store_compressed_length(buf, typelib->type_lengths[i]);
        buf.append(typelib->type_names[i], typelib->type_lengths[i]);
      }
    }
  }
  if (buf.length() > 0)
    return write_tlv_field(m_metadata_buf, SET_STR_VALUE, buf);
  return false;
}

bool Table_map_log_event::init_enum_str_value_field()
{
  StringBuffer<1024> buf;
  TYPELIB *typelib;

  /* ENUM is same to SET columns, see comment in init_set_str_value_field */
  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    if ((typelib= binlog_type_info_array[i].m_enum_typelib))
    {
      store_compressed_length(buf, typelib->count);
      for (unsigned int i= 0; i < typelib->count; i++)
      {
        store_compressed_length(buf, typelib->type_lengths[i]);
        buf.append(typelib->type_names[i], typelib->type_lengths[i]);
      }
    }
  }

  if (buf.length() > 0)
    return write_tlv_field(m_metadata_buf, ENUM_STR_VALUE, buf);
  return false;
}

bool Table_map_log_event::init_geometry_type_field()
{
  StringBuffer<256> buf;
  uint geom_type;

  /* Geometry type of geometry columns is stored one by one as packed length */
  for (unsigned int i= 0 ; i < m_table->s->fields ; ++i)
  {
    if (binlog_type_info_array[i].m_type_code == MYSQL_TYPE_GEOMETRY)
    {
      geom_type= binlog_type_info_array[i].m_geom_type;
      DBUG_EXECUTE_IF("inject_invalid_geometry_type", geom_type= 100;);
      store_compressed_length(buf, geom_type);
    }
  }

  if (buf.length() > 0)
    return write_tlv_field(m_metadata_buf, GEOMETRY_TYPE, buf);
  return false;
}

bool Table_map_log_event::init_primary_key_field()
{
  DBUG_EXECUTE_IF("simulate_init_primary_key_field_error", return true;);

  if (unlikely(m_table->s->primary_key == MAX_KEY))
    return false;

  // If any key column uses prefix like KEY(c1(10)) */
  bool has_prefix= false;
  KEY *pk= m_table->key_info + m_table->s->primary_key;

  DBUG_ASSERT(pk->user_defined_key_parts > 0);

  /* Check if any key column uses prefix */
  for (uint i= 0; i < pk->user_defined_key_parts; i++)
  {
    KEY_PART_INFO *key_part= pk->key_part+i;
    if (key_part->length != m_table->field[key_part->fieldnr-1]->key_length())
    {
      has_prefix= true;
      break;
    }
  }

  StringBuffer<128> buf;

  if (!has_prefix)
  {
    /* Index of PK columns are stored one by one. */
    for (uint i= 0; i < pk->user_defined_key_parts; i++)
    {
      KEY_PART_INFO *key_part= pk->key_part+i;
      store_compressed_length(buf, key_part->fieldnr-1);
    }
    return write_tlv_field(m_metadata_buf, SIMPLE_PRIMARY_KEY, buf);
  }
  else
  {
    /* Index of PK columns are stored with a prefix length one by one. */
    for (uint i= 0; i < pk->user_defined_key_parts; i++)
    {
      KEY_PART_INFO *key_part= pk->key_part+i;
      size_t prefix= 0;

      store_compressed_length(buf, key_part->fieldnr-1);

      // Store character length but not octet length
      if (key_part->length != m_table->field[key_part->fieldnr-1]->key_length())
        prefix= key_part->length / key_part->field->charset()->mbmaxlen;
      store_compressed_length(buf, prefix);
    }
    return write_tlv_field(m_metadata_buf, PRIMARY_KEY_WITH_PREFIX, buf);
  }
}
7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116

#if defined(HAVE_REPLICATION)
/*
  Print some useful information for the SHOW BINARY LOG information
  field.
 */

void Table_map_log_event::pack_info(Protocol *protocol)
{
    char buf[256];
    size_t bytes= my_snprintf(buf, sizeof(buf),
7117
                              "table_id: %llu (%s.%s)",
7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131
                              m_table_id, m_dbnam, m_tblnam);
    protocol->store(buf, bytes, &my_charset_bin);
}
#endif


/**************************************************************************
	Write_rows_log_event member functions
**************************************************************************/

/*
  Constructor used to build an event for writing to the binary log.
 */
Write_rows_log_event::Write_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
7132
                                           ulonglong tid_arg,
7133 7134 7135 7136 7137 7138 7139 7140 7141
                                           bool is_transactional)
  :Rows_log_event(thd_arg, tbl_arg, tid_arg, tbl_arg->rpl_write_set,
                  is_transactional, WRITE_ROWS_EVENT_V1)
{
}

Write_rows_compressed_log_event::Write_rows_compressed_log_event(
                                           THD *thd_arg,
                                           TABLE *tbl_arg,
7142
                                           ulonglong tid_arg,
7143 7144 7145 7146 7147 7148 7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173 7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196
                                           bool is_transactional)
  : Write_rows_log_event(thd_arg, tbl_arg, tid_arg, is_transactional)
{
  m_type = WRITE_ROWS_COMPRESSED_EVENT_V1;
}

bool Write_rows_compressed_log_event::write()
{
  return Rows_log_event::write_compressed();
}


#if defined(HAVE_REPLICATION)
int 
Write_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
  int error= 0;

  /*
    Increment the global status insert count variable
  */
  if (get_flags(STMT_END_F))
    status_var_increment(thd->status_var.com_stat[SQLCOM_INSERT]);

  /**
     todo: to introduce a property for the event (handler?) which forces
     applying the event in the replace (idempotent) fashion.
  */
  if (slave_exec_mode == SLAVE_EXEC_MODE_IDEMPOTENT)
  {
    /*
      We are using REPLACE semantics and not INSERT IGNORE semantics
      when writing rows, that is: new rows replace old rows.  We need to
      inform the storage engine that it should use this behaviour.
    */
    
    /* Tell the storage engine that we are using REPLACE semantics. */
    thd->lex->duplicates= DUP_REPLACE;
    
    /*
      Pretend we're executing a REPLACE command: this is needed for
      InnoDB since it is not (properly) checking the lex->duplicates flag.
    */
    thd->lex->sql_command= SQLCOM_REPLACE;
    /* 
       Do not raise the error flag in case of hitting to an unique attribute
    */
    m_table->file->extra(HA_EXTRA_IGNORE_DUP_KEY);
    /* 
       The following is needed in case if we have AFTER DELETE triggers.
    */
    m_table->file->extra(HA_EXTRA_WRITE_CAN_REPLACE);
    m_table->file->extra(HA_EXTRA_IGNORE_NO_KEY);
  }
7197
  if (m_table->triggers && do_invoke_trigger())
7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212 7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223 7224 7225 7226 7227
    m_table->prepare_triggers_for_insert_stmt_or_event();

  /* Honor next number column if present */
  m_table->next_number_field= m_table->found_next_number_field;
  /*
   * Fixed Bug#45999, In RBR, Store engine of Slave auto-generates new
   * sequence numbers for auto_increment fields if the values of them are 0.
   * If generateing a sequence number is decided by the values of
   * table->auto_increment_field_not_null and SQL_MODE(if includes
   * MODE_NO_AUTO_VALUE_ON_ZERO) in update_auto_increment function.
   * SQL_MODE of slave sql thread is always consistency with master's.
   * In RBR, auto_increment fields never are NULL, except if the auto_inc
   * column exists only on the slave side (i.e., in an extra column
   * on the slave's table).
   */
  if (!is_auto_inc_in_extra_columns())
    m_table->auto_increment_field_not_null= TRUE;
  else
  {
    /*
      Here we have checked that there is an extra field
      on this server's table that has an auto_inc column.

      Mark that the auto_increment field is null and mark
      the read and write set bits.

      (There can only be one AUTO_INC column, it is always
       indexed and it cannot have a DEFAULT value).
    */
    m_table->auto_increment_field_not_null= FALSE;
7228
    m_table->mark_auto_increment_column(true);
7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277 7278 7279 7280 7281 7282 7283 7284
  }

  return error;
}

int 
Write_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const,
                                              int error)
{
  int local_error= 0;

  /**
    Clear the write_set bit for auto_inc field that only
    existed on the destination table as an extra column.
   */
  if (is_auto_inc_in_extra_columns())
  {
    bitmap_clear_bit(m_table->rpl_write_set,
                     m_table->next_number_field->field_index);
    bitmap_clear_bit(m_table->read_set,
                     m_table->next_number_field->field_index);

    if (get_flags(STMT_END_F))
      m_table->file->ha_release_auto_increment();
  }
  m_table->next_number_field=0;
  m_table->auto_increment_field_not_null= FALSE;
  if (slave_exec_mode == SLAVE_EXEC_MODE_IDEMPOTENT)
  {
    m_table->file->extra(HA_EXTRA_NO_IGNORE_DUP_KEY);
    m_table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
    /*
      resetting the extra with 
      table->file->extra(HA_EXTRA_NO_IGNORE_NO_KEY); 
      fires bug#27077
      explanation: file->reset() performs this duty
      ultimately. Still todo: fix
    */
  }
  if (unlikely((local_error= m_table->file->ha_end_bulk_insert())))
  {
    m_table->file->print_error(local_error, MYF(0));
  }
  return error? error : local_error;
}

bool Rows_log_event::process_triggers(trg_event_type event,
                                      trg_action_time_type time_type,
                                      bool old_row_is_record1)
{
  bool result;
  DBUG_ENTER("Rows_log_event::process_triggers");
  m_table->triggers->mark_fields_used(event);
  if (slave_run_triggers_for_rbr == SLAVE_RUN_TRIGGERS_FOR_RBR_YES)
  {
    result= m_table->triggers->process_triggers(thd, event,
7285 7286
                                                time_type,
                                                old_row_is_record1);
7287 7288 7289
  }
  else
    result= m_table->triggers->process_triggers(thd, event,
7290 7291
                                                time_type,
                                                old_row_is_record1);
7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356 7357 7358 7359 7360 7361 7362 7363 7364 7365 7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376

  DBUG_RETURN(result);
}
/*
  Check if there are more UNIQUE keys after the given key.
*/
static int
last_uniq_key(TABLE *table, uint keyno)
{
  while (++keyno < table->s->keys)
    if (table->key_info[keyno].flags & HA_NOSAME)
      return 0;
  return 1;
}

/**
   Check if an error is a duplicate key error.

   This function is used to check if an error code is one of the
   duplicate key error, i.e., and error code for which it is sensible
   to do a <code>get_dup_key()</code> to retrieve the duplicate key.

   @param errcode The error code to check.

   @return <code>true</code> if the error code is such that
   <code>get_dup_key()</code> will return true, <code>false</code>
   otherwise.
 */
bool
is_duplicate_key_error(int errcode)
{
  switch (errcode)
  {
  case HA_ERR_FOUND_DUPP_KEY:
  case HA_ERR_FOUND_DUPP_UNIQUE:
    return true;
  }
  return false;
}

/**
  Write the current row into event's table.

  The row is located in the row buffer, pointed by @c m_curr_row member.
  Number of columns of the row is stored in @c m_width member (it can be 
  different from the number of columns in the table to which we insert). 
  Bitmap @c m_cols indicates which columns are present in the row. It is assumed 
  that event's table is already open and pointed by @c m_table.

  If the same record already exists in the table it can be either overwritten 
  or an error is reported depending on the value of @c overwrite flag 
  (error reporting not yet implemented). Note that the matching record can be
  different from the row we insert if we use primary keys to identify records in
  the table.

  The row to be inserted can contain values only for selected columns. The 
  missing columns are filled with default values using @c prepare_record() 
  function. If a matching record is found in the table and @c overwritte is
  true, the missing columns are taken from it.

  @param  rli   Relay log info (needed for row unpacking).
  @param  overwrite  
                Shall we overwrite if the row already exists or signal 
                error (currently ignored).

  @returns Error code on failure, 0 on success.

  This method, if successful, sets @c m_curr_row_end pointer to point at the
  next row in the rows buffer. This is done when unpacking the row to be 
  inserted.

  @note If a matching record is found, it is either updated using 
  @c ha_update_row() or first deleted and then new record written.
*/ 

int
Rows_log_event::write_row(rpl_group_info *rgi,
                          const bool overwrite)
{
  DBUG_ENTER("write_row");
  DBUG_ASSERT(m_table != NULL && thd != NULL);

  TABLE *table= m_table;  // pointer to event's table
  int error;
  int UNINIT_VAR(keynum);
7377
  const bool invoke_triggers= (m_table->triggers && do_invoke_trigger());
7378 7379 7380 7381 7382 7383 7384 7385 7386 7387 7388
  auto_afree_ptr<char> key(NULL);

  prepare_record(table, m_width, true);

  /* unpack row into table->record[0] */
  if (unlikely((error= unpack_current_row(rgi))))
  {
    table->file->print_error(error, MYF(0));
    DBUG_RETURN(error);
  }

Marko Mäkelä's avatar
Marko Mäkelä committed
7389
  if (m_curr_row == m_rows_buf && !invoke_triggers && !table->s->long_unique_table)
7390 7391 7392 7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403 7404 7405 7406 7407 7408 7409 7410 7411 7412 7413 7414 7415 7416 7417 7418 7419 7420 7421
  {
    /*
       This table has no triggers so we can do bulk insert.

       This is the first row to be inserted, we estimate the rows with
       the size of the first row and use that value to initialize
       storage engine for bulk insertion.
    */
    /* this is the first row to be inserted, we estimate the rows with
       the size of the first row and use that value to initialize
       storage engine for bulk insertion */
    DBUG_ASSERT(!(m_curr_row > m_curr_row_end));
    ha_rows estimated_rows= 0;
    if (m_curr_row < m_curr_row_end)
      estimated_rows= (m_rows_end - m_curr_row) / (m_curr_row_end - m_curr_row);
    else if (m_curr_row == m_curr_row_end)
      estimated_rows= 1;

    table->file->ha_start_bulk_insert(estimated_rows);
  }

  /*
    Explicitly set the auto_inc to null to make sure that
    it gets an auto_generated value.
  */
  if (is_auto_inc_in_extra_columns())
    m_table->next_number_field->set_null();
  
  DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
  DBUG_PRINT_BITSET("debug", "rpl_write_set: %s", table->rpl_write_set);
  DBUG_PRINT_BITSET("debug", "read_set:      %s", table->read_set);

Marko Mäkelä's avatar
Marko Mäkelä committed
7422 7423 7424
  if (table->s->long_unique_table)
    table->update_virtual_fields(table->file, VCOL_UPDATE_FOR_WRITE);

7425 7426 7427 7428 7429 7430 7431 7432 7433 7434 7435
  if (invoke_triggers &&
      unlikely(process_triggers(TRG_EVENT_INSERT, TRG_ACTION_BEFORE, TRUE)))
  {
    DBUG_RETURN(HA_ERR_GENERIC); // in case if error is not set yet
  }

  // Handle INSERT.
  if (table->versioned(VERS_TIMESTAMP))
  {
    ulong sec_part;
    bitmap_set_bit(table->read_set, table->vers_start_field()->field_index);
Marko Mäkelä's avatar
Marko Mäkelä committed
7436
    table->file->column_bitmaps_signal();
7437 7438 7439 7440 7441 7442 7443 7444 7445 7446 7447 7448 7449
    // Check whether a row came from unversioned table and fix vers fields.
    if (table->vers_start_field()->get_timestamp(&sec_part) == 0 && sec_part == 0)
      table->vers_update_fields();
  }

  /* 
    Try to write record. If a corresponding record already exists in the table,
    we try to change it using ha_update_row() if possible. Otherwise we delete
    it and repeat the whole process again. 

    TODO: Add safety measures against infinite looping. 
   */

7450 7451
  DBUG_EXECUTE_IF("write_row_inject_sleep_before_ha_write_row",
                  my_sleep(20000););
7452 7453 7454 7455 7456 7457 7458 7459 7460 7461 7462 7463 7464 7465 7466 7467 7468 7469 7470 7471 7472 7473 7474 7475 7476 7477 7478 7479 7480 7481 7482 7483 7484 7485 7486 7487 7488 7489 7490 7491 7492 7493 7494 7495 7496 7497 7498 7499 7500 7501 7502 7503 7504 7505 7506 7507 7508 7509 7510 7511 7512 7513 7514 7515 7516 7517 7518 7519 7520 7521 7522 7523 7524 7525 7526 7527 7528 7529 7530 7531 7532 7533 7534 7535 7536 7537
  if (table->s->sequence)
    error= update_sequence();
  else while (unlikely(error= table->file->ha_write_row(table->record[0])))
  {
    if (error == HA_ERR_LOCK_DEADLOCK ||
        error == HA_ERR_LOCK_WAIT_TIMEOUT ||
        (keynum= table->file->get_dup_key(error)) < 0 ||
        !overwrite)
    {
      DBUG_PRINT("info",("get_dup_key returns %d)", keynum));
      /*
        Deadlock, waiting for lock or just an error from the handler
        such as HA_ERR_FOUND_DUPP_KEY when overwrite is false.
        Retrieval of the duplicate key number may fail
        - either because the error was not "duplicate key" error
        - or because the information which key is not available
      */
      table->file->print_error(error, MYF(0));
      DBUG_RETURN(error);
    }
    /*
       We need to retrieve the old row into record[1] to be able to
       either update or delete the offending record.  We either:

       - use rnd_pos() with a row-id (available as dupp_row) to the
         offending row, if that is possible (MyISAM and Blackhole), or else

       - use index_read_idx() with the key that is duplicated, to
         retrieve the offending row.
     */
    if (table->file->ha_table_flags() & HA_DUPLICATE_POS)
    {
      DBUG_PRINT("info",("Locating offending record using rnd_pos()"));

      if ((error= table->file->ha_rnd_init_with_error(0)))
      {
        DBUG_RETURN(error);
      }

      error= table->file->ha_rnd_pos(table->record[1], table->file->dup_ref);
      if (unlikely(error))
      {
        DBUG_PRINT("info",("rnd_pos() returns error %d",error));
        table->file->print_error(error, MYF(0));
        DBUG_RETURN(error);
      }
      table->file->ha_rnd_end();
    }
    else
    {
      DBUG_PRINT("info",("Locating offending record using index_read_idx()"));

      if (table->file->extra(HA_EXTRA_FLUSH_CACHE))
      {
        DBUG_PRINT("info",("Error when setting HA_EXTRA_FLUSH_CACHE"));
        DBUG_RETURN(my_errno);
      }

      if (key.get() == NULL)
      {
        key.assign(static_cast<char*>(my_alloca(table->s->max_unique_length)));
        if (key.get() == NULL)
        {
          DBUG_PRINT("info",("Can't allocate key buffer"));
          DBUG_RETURN(ENOMEM);
        }
      }

      key_copy((uchar*)key.get(), table->record[0], table->key_info + keynum,
               0);
      error= table->file->ha_index_read_idx_map(table->record[1], keynum,
                                                (const uchar*)key.get(),
                                                HA_WHOLE_KEY,
                                                HA_READ_KEY_EXACT);
      if (unlikely(error))
      {
        DBUG_PRINT("info",("index_read_idx() returns %s", HA_ERR(error)));
        table->file->print_error(error, MYF(0));
        DBUG_RETURN(error);
      }
    }

    /*
       Now, record[1] should contain the offending row.  That
       will enable us to update it or, alternatively, delete it (so
       that we can insert the new row afterwards).
Marko Mäkelä's avatar
Marko Mäkelä committed
7538 7539 7540 7541 7542 7543 7544 7545
    */
    if (table->s->long_unique_table)
    {
      /* same as for REPLACE/ODKU */
      table->move_fields(table->field, table->record[1], table->record[0]);
      table->update_virtual_fields(table->file, VCOL_UPDATE_FOR_REPLACE);
      table->move_fields(table->field, table->record[0], table->record[1]);
    }
7546 7547 7548 7549 7550 7551 7552 7553 7554

    /*
      If row is incomplete we will use the record found to fill 
      missing columns.  
    */
    if (!get_flags(COMPLETE_ROWS_F))
    {
      restore_record(table,record[1]);
      error= unpack_current_row(rgi);
Marko Mäkelä's avatar
Marko Mäkelä committed
7555 7556
      if (table->s->long_unique_table)
        table->update_virtual_fields(table->file, VCOL_UPDATE_FOR_WRITE);
7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569 7570 7571 7572 7573 7574 7575 7576 7577 7578 7579 7580 7581 7582 7583 7584 7585 7586 7587 7588 7589 7590 7591 7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602 7603 7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621 7622 7623 7624 7625 7626 7627 7628 7629 7630 7631 7632 7633 7634 7635 7636 7637 7638
    }

    DBUG_PRINT("debug",("preparing for update: before and after image"));
    DBUG_DUMP("record[1] (before)", table->record[1], table->s->reclength);
    DBUG_DUMP("record[0] (after)", table->record[0], table->s->reclength);

    /*
       REPLACE is defined as either INSERT or DELETE + INSERT.  If
       possible, we can replace it with an UPDATE, but that will not
       work on InnoDB if FOREIGN KEY checks are necessary.

       I (Matz) am not sure of the reason for the last_uniq_key()
       check as, but I'm guessing that it's something along the
       following lines.

       Suppose that we got the duplicate key to be a key that is not
       the last unique key for the table and we perform an update:
       then there might be another key for which the unique check will
       fail, so we're better off just deleting the row and inserting
       the correct row.

       Additionally we don't use UPDATE if rbr triggers should be invoked -
       when triggers are used we want a simple and predictable execution path.
     */
    if (last_uniq_key(table, keynum) && !invoke_triggers &&
        !table->file->referenced_by_foreign_key())
    {
      DBUG_PRINT("info",("Updating row using ha_update_row()"));
      error= table->file->ha_update_row(table->record[1],
                                       table->record[0]);
      switch (error) {

      case HA_ERR_RECORD_IS_THE_SAME:
        DBUG_PRINT("info",("ignoring HA_ERR_RECORD_IS_THE_SAME error from"
                           " ha_update_row()"));
        error= 0;

      case 0:
        break;

      default:
        DBUG_PRINT("info",("ha_update_row() returns error %d",error));
        table->file->print_error(error, MYF(0));
      }

      DBUG_RETURN(error);
    }
    else
    {
      DBUG_PRINT("info",("Deleting offending row and trying to write new one again"));
      if (invoke_triggers &&
          unlikely(process_triggers(TRG_EVENT_DELETE, TRG_ACTION_BEFORE,
                                    TRUE)))
        error= HA_ERR_GENERIC; // in case if error is not set yet
      else
      {
        if (unlikely((error= table->file->ha_delete_row(table->record[1]))))
        {
          DBUG_PRINT("info",("ha_delete_row() returns error %d",error));
          table->file->print_error(error, MYF(0));
          DBUG_RETURN(error);
        }
        if (invoke_triggers &&
            unlikely(process_triggers(TRG_EVENT_DELETE, TRG_ACTION_AFTER,
                                      TRUE)))
          DBUG_RETURN(HA_ERR_GENERIC); // in case if error is not set yet
      }
      /* Will retry ha_write_row() with the offending row removed. */
    }
  }

  if (invoke_triggers &&
      unlikely(process_triggers(TRG_EVENT_INSERT, TRG_ACTION_AFTER, TRUE)))
    error= HA_ERR_GENERIC; // in case if error is not set yet

  DBUG_RETURN(error);
}


int Rows_log_event::update_sequence()
{
  TABLE *table= m_table;  // pointer to event's table
7639 7640
  bool old_master= false;
  int err= 0;
7641

7642
  if (!bitmap_is_set(table->rpl_write_set, MIN_VALUE_FIELD_NO) ||
7643 7644 7645 7646
      (
#if defined(WITH_WSREP)
       ! WSREP(thd) &&
#endif
7647
       table->in_use->rgi_slave &&
7648
       !(table->in_use->rgi_slave->gtid_ev_flags2 & Gtid_log_event::FL_DDL) &&
7649 7650 7651
       !(old_master=
         rpl_master_has_bug(thd->rgi_slave->rli,
                            29621, FALSE, FALSE, FALSE, TRUE))))
7652 7653 7654 7655
  {
    /* This event come from a setval function executed on the master.
       Update the sequence next_number and round, like we do with setval()
    */
7656 7657
    MY_BITMAP *old_map= dbug_tmp_use_all_columns(table,
                                                 &table->read_set);
7658 7659
    longlong nextval= table->field[NEXT_FIELD_NO]->val_int();
    longlong round= table->field[ROUND_FIELD_NO]->val_int();
7660
    dbug_tmp_restore_column_map(&table->read_set, old_map);
7661 7662 7663

    return table->s->sequence->set_value(table, nextval, round, 0) > 0;
  }
7664
  if (old_master && !WSREP(thd) && thd->rgi_slave->is_parallel_exec)
7665 7666 7667 7668 7669 7670 7671 7672 7673 7674 7675 7676 7677 7678 7679
  {
    DBUG_ASSERT(thd->rgi_slave->parallel_entry);
    /*
      With parallel replication enabled, we can't execute alongside any other
      transaction in which we may depend, so we force retry to release
      the server layer table lock for possible prior in binlog order
      same table transactions.
    */
    if (thd->rgi_slave->parallel_entry->last_committed_sub_id <
        thd->rgi_slave->wait_commit_sub_id)
    {
      err= ER_LOCK_DEADLOCK;
      my_error(err, MYF(0));
    }
  }
7680 7681 7682 7683
  /*
    Update all fields in table and update the active sequence, like with
    ALTER SEQUENCE
  */
7684
  return err == 0 ? table->file->ha_write_row(table->record[0]) : err;
7685 7686 7687 7688 7689 7690 7691 7692 7693 7694 7695 7696 7697
}


#endif


#if defined(HAVE_REPLICATION)

int
Write_rows_log_event::do_exec_row(rpl_group_info *rgi)
{
  DBUG_ASSERT(m_table != NULL);
  const char *tmp= thd->get_proc_info();
7698
  char *message, msg[128];
Marko Mäkelä's avatar
Marko Mäkelä committed
7699 7700 7701 7702 7703 7704
  const LEX_CSTRING &table_name= m_table->s->table_name;
  const char quote_char=
    get_quote_char_for_identifier(thd, table_name.str, table_name.length);
  my_snprintf(msg, sizeof msg,
              "Write_rows_log_event::write_row() on table %c%.*s%c",
              quote_char, int(table_name.length), table_name.str, quote_char);
7705
  message= msg;
7706 7707 7708 7709
  int error;

#ifdef WSREP_PROC_INFO
  my_snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
7710 7711 7712
              "Write_rows_log_event::write_row(%lld) on table %c%.*s%c",
              (long long) wsrep_thd_trx_seqno(thd), quote_char,
              int(table_name.length), table_name.str, quote_char);
7713 7714 7715 7716 7717 7718 7719 7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743 7744 7745 7746 7747 7748 7749
  message= thd->wsrep_info;
#endif /* WSREP_PROC_INFO */

  thd_proc_info(thd, message);
  error= write_row(rgi, slave_exec_mode == SLAVE_EXEC_MODE_IDEMPOTENT);
  thd_proc_info(thd, tmp);

  if (unlikely(error) && unlikely(!thd->is_error()))
  {
    DBUG_ASSERT(0);
    my_error(ER_UNKNOWN_ERROR, MYF(0));
  }

  return error;
}

#endif /* defined(HAVE_REPLICATION) */


#if defined(HAVE_REPLICATION)
uint8 Write_rows_log_event::get_trg_event_map()
{
  return trg2bit(TRG_EVENT_INSERT) | trg2bit(TRG_EVENT_UPDATE) |
         trg2bit(TRG_EVENT_DELETE);
}
#endif

/**************************************************************************
	Delete_rows_log_event member functions
**************************************************************************/

#if defined(HAVE_REPLICATION)
/*
  Compares table->record[0] and table->record[1]

  Returns TRUE if different.
*/
Marko Mäkelä's avatar
Marko Mäkelä committed
7750
static bool record_compare(TABLE *table, bool vers_from_plain= false)
7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762 7763 7764 7765 7766 7767 7768 7769 7770 7771 7772 7773 7774 7775 7776 7777 7778 7779 7780 7781 7782
{
  bool result= FALSE;
  /**
    Compare full record only if:
    - there are no blob fields (otherwise we would also need 
      to compare blobs contents as well);
    - there are no varchar fields (otherwise we would also need
      to compare varchar contents as well);
    - there are no null fields, otherwise NULLed fields 
      contents (i.e., the don't care bytes) may show arbitrary 
      values, depending on how each engine handles internally.
    */
  if ((table->s->blob_fields + 
       table->s->varchar_fields + 
       table->s->null_fields) == 0)
  {
    result= cmp_record(table,record[1]);
    goto record_compare_exit;
  }

  /* Compare null bits */
  if (memcmp(table->null_flags,
	     table->null_flags+table->s->rec_buff_length,
	     table->s->null_bytes))
  {
    result= TRUE;				// Diff in NULL value
    goto record_compare_exit;
  }

  /* Compare fields */
  for (Field **ptr=table->field ; *ptr ; ptr++)
  {
Marko Mäkelä's avatar
Marko Mäkelä committed
7783 7784 7785 7786 7787 7788 7789 7790 7791 7792 7793 7794
    /*
      If the table is versioned, don't compare using the version if there is a
      primary key. If there isn't a primary key, we need the version to
      identify the correct record if there are duplicate rows in the data set.
      However, if the primary server is unversioned (vers_from_plain is true),
      then we implicitly use row_end as the primary key on our side. This is
      because the implicit row_end value will be set to the maximum value for
      the latest row update (which is what we care about).
    */
    if (table->versioned() && (*ptr)->vers_sys_field() &&
        (table->s->primary_key < MAX_KEY ||
         (vers_from_plain && table->vers_start_field() == (*ptr))))
7795 7796 7797 7798 7799 7800 7801 7802 7803 7804 7805 7806 7807 7808 7809 7810 7811 7812 7813 7814 7815 7816 7817 7818 7819 7820 7821 7822 7823 7824 7825 7826 7827 7828 7829 7830 7831 7832 7833 7834 7835 7836 7837 7838 7839 7840 7841 7842 7843 7844 7845 7846 7847 7848 7849 7850 7851 7852 7853 7854 7855 7856 7857 7858 7859 7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875 7876 7877 7878 7879 7880 7881 7882
      continue;
    /**
      We only compare field contents that are not null.
      NULL fields (i.e., their null bits) were compared 
      earlier.
    */
    if (!(*(ptr))->is_null())
    {
      if ((*ptr)->cmp_binary_offset(table->s->rec_buff_length))
      {
        result= TRUE;
        goto record_compare_exit;
      }
    }
  }

record_compare_exit:
  return result;
}


/**
  Find the best key to use when locating the row in @c find_row().

  A primary key is preferred if it exists; otherwise a unique index is
  preferred. Else we pick the index with the smalles rec_per_key value.

  If a suitable key is found, set @c m_key, @c m_key_nr and @c m_key_info
  member fields appropriately.

  @returns Error code on failure, 0 on success.
*/
int Rows_log_event::find_key()
{
  uint i, best_key_nr, last_part;
  KEY *key, *UNINIT_VAR(best_key);
  ulong UNINIT_VAR(best_rec_per_key), tmp;
  DBUG_ENTER("Rows_log_event::find_key");
  DBUG_ASSERT(m_table);

  best_key_nr= MAX_KEY;

  /*
    Keys are sorted so that any primary key is first, followed by unique keys,
    followed by any other. So we will automatically pick the primary key if
    it exists.
  */
  for (i= 0, key= m_table->key_info; i < m_table->s->keys; i++, key++)
  {
    if (!m_table->s->keys_in_use.is_set(i))
      continue;
    /*
      We cannot use a unique key with NULL-able columns to uniquely identify
      a row (but we can still select it for range scan below if nothing better
      is available).
    */
    if ((key->flags & (HA_NOSAME | HA_NULL_PART_KEY)) == HA_NOSAME)
    {
      best_key_nr= i;
      best_key= key;
      break;
    }
    /*
      We can only use a non-unique key if it allows range scans (ie. skip
      FULLTEXT indexes and such).
    */
    last_part= key->user_defined_key_parts - 1;
    DBUG_PRINT("info", ("Index %s rec_per_key[%u]= %lu",
                        key->name.str, last_part, key->rec_per_key[last_part]));
    if (!(m_table->file->index_flags(i, last_part, 1) & HA_READ_NEXT))
      continue;

    tmp= key->rec_per_key[last_part];
    if (best_key_nr == MAX_KEY || (tmp > 0 && tmp < best_rec_per_key))
    {
      best_key_nr= i;
      best_key= key;
      best_rec_per_key= tmp;
    }
  }

  if (best_key_nr == MAX_KEY)
  {
    m_key_info= NULL;
    DBUG_RETURN(0);
  }

  // Allocate buffer for key searches
7883
  m_key= (uchar *) my_malloc(PSI_INSTRUMENT_ME, best_key->key_length, MYF(MY_WME));
7884 7885 7886 7887 7888 7889 7890 7891 7892 7893 7894 7895 7896 7897 7898 7899 7900 7901 7902 7903 7904 7905 7906 7907 7908 7909 7910 7911 7912 7913 7914 7915 7916 7917 7918 7919 7920 7921 7922 7923 7924 7925 7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937 7938 7939 7940 7941 7942 7943 7944 7945 7946 7947 7948 7949
  if (m_key == NULL)
    DBUG_RETURN(HA_ERR_OUT_OF_MEM);
  m_key_info= best_key;
  m_key_nr= best_key_nr;

  DBUG_RETURN(0);;
}


/* 
  Check if we are already spending too much time on this statement.
  if we are, warn user that it might be because table does not have
  a PK, but only if the warning was not printed before for this STMT.

  @param type          The event type code.
  @param table_name    The name of the table that the slave is 
                       operating.
  @param is_index_scan States whether the slave is doing an index scan 
                       or not.
  @param rli           The relay metadata info.
*/
static inline 
void issue_long_find_row_warning(Log_event_type type, 
                                 const char *table_name,
                                 bool is_index_scan,
                                 rpl_group_info *rgi)
{
  if ((global_system_variables.log_warnings > 1 && 
       !rgi->is_long_find_row_note_printed()))
  {
    ulonglong now= microsecond_interval_timer();
    ulonglong stmt_ts= rgi->get_row_stmt_start_timestamp();
    
    DBUG_EXECUTE_IF("inject_long_find_row_note", 
                    stmt_ts-=(LONG_FIND_ROW_THRESHOLD*2*HRTIME_RESOLUTION););

    longlong delta= (now - stmt_ts)/HRTIME_RESOLUTION;

    if (delta > LONG_FIND_ROW_THRESHOLD)
    {
      rgi->set_long_find_row_note_printed();
      const char* evt_type= LOG_EVENT_IS_DELETE_ROW(type) ? " DELETE" : "n UPDATE";
      const char* scan_type= is_index_scan ? "scanning an index" : "scanning the table";

      sql_print_information("The slave is applying a ROW event on behalf of a%s statement "
                            "on table %s and is currently taking a considerable amount "
                            "of time (%lld seconds). This is due to the fact that it is %s "
                            "while looking up records to be processed. Consider adding a "
                            "primary key (or unique key) to the table to improve "
                            "performance.",
                            evt_type, table_name, (long) delta, scan_type);
    }
  }
}


/*
  HA_ERR_KEY_NOT_FOUND is a fatal error normally, but it's an expected
  error in speculate optimistic mode, so use something non-fatal instead
*/
static int row_not_found_error(rpl_group_info *rgi)
{
  return rgi->speculation != rpl_group_info::SPECULATE_OPTIMISTIC
         ? HA_ERR_KEY_NOT_FOUND : HA_ERR_RECORD_CHANGED;
}

7950 7951 7952 7953 7954 7955
static int end_of_file_error(rpl_group_info *rgi)
{
  return rgi->speculation != rpl_group_info::SPECULATE_OPTIMISTIC
         ? HA_ERR_END_OF_FILE : HA_ERR_RECORD_CHANGED;
}

7956 7957 7958 7959 7960 7961 7962 7963 7964 7965 7966 7967 7968 7969 7970 7971 7972 7973 7974 7975 7976 7977 7978 7979 7980 7981 7982 7983 7984 7985 7986 7987 7988 7989 7990 7991 7992 7993 7994 7995 7996 7997 7998 7999 8000 8001 8002 8003 8004 8005 8006 8007 8008 8009 8010 8011 8012 8013 8014 8015 8016 8017 8018 8019 8020 8021 8022
/**
  Locate the current row in event's table.

  The current row is pointed by @c m_curr_row. Member @c m_width tells
  how many columns are there in the row (this can be differnet from
  the number of columns in the table). It is assumed that event's
  table is already open and pointed by @c m_table.

  If a corresponding record is found in the table it is stored in 
  @c m_table->record[0]. Note that when record is located based on a primary 
  key, it is possible that the record found differs from the row being located.

  If no key is specified or table does not have keys, a table scan is used to 
  find the row. In that case the row should be complete and contain values for
  all columns. However, it can still be shorter than the table, i.e. the table 
  can contain extra columns not present in the row. It is also possible that 
  the table has fewer columns than the row being located. 

  @returns Error code on failure, 0 on success. 
  
  @post In case of success @c m_table->record[0] contains the record found. 
  Also, the internal "cursor" of the table is positioned at the record found.

  @note If the engine allows random access of the records, a combination of
  @c position() and @c rnd_pos() will be used. 

  Note that one MUST call ha_index_or_rnd_end() after this function if
  it returns 0 as we must leave the row position in the handler intact
  for any following update/delete command.
*/

int Rows_log_event::find_row(rpl_group_info *rgi)
{
  DBUG_ENTER("Rows_log_event::find_row");

  DBUG_ASSERT(m_table && m_table->in_use != NULL);

  TABLE *table= m_table;
  int error= 0;
  bool is_table_scan= false, is_index_scan= false;

  /*
    rpl_row_tabledefs.test specifies that
    if the extra field on the slave does not have a default value
    and this is okay with Delete or Update events.
    Todo: fix wl3228 hld that requires defauls for all types of events
  */
  
  prepare_record(table, m_width, FALSE);
  error= unpack_current_row(rgi);

  m_vers_from_plain= false;
  if (table->versioned())
  {
    Field *row_end= table->vers_end_field();
    DBUG_ASSERT(table->read_set);
    bitmap_set_bit(table->read_set, row_end->field_index);
    // check whether master table is unversioned
    if (row_end->val_int() == 0)
    {
      bitmap_set_bit(table->write_set, row_end->field_index);
      // Plain source table may have a PRIMARY KEY. And row_end is always
      // a part of PRIMARY KEY. Set it to max value for engine to find it in
      // index. Needed for an UPDATE/DELETE cases.
      table->vers_end_field()->set_max();
      m_vers_from_plain= true;
    }
Marko Mäkelä's avatar
Marko Mäkelä committed
8023
    table->file->column_bitmaps_signal();
8024 8025 8026 8027 8028 8029 8030 8031 8032 8033 8034 8035 8036 8037 8038 8039 8040 8041 8042 8043 8044 8045 8046 8047 8048 8049 8050 8051 8052 8053 8054 8055 8056 8057 8058 8059 8060 8061 8062 8063 8064 8065 8066 8067 8068 8069 8070 8071 8072 8073 8074 8075 8076 8077 8078 8079 8080 8081 8082 8083 8084 8085 8086 8087 8088 8089 8090 8091 8092 8093 8094 8095 8096 8097 8098 8099 8100 8101 8102 8103 8104 8105 8106 8107 8108 8109 8110 8111 8112 8113 8114 8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126 8127 8128 8129 8130 8131 8132 8133 8134 8135 8136 8137 8138 8139 8140 8141 8142 8143 8144 8145 8146 8147 8148 8149 8150 8151 8152 8153 8154 8155 8156 8157 8158 8159 8160 8161 8162 8163 8164 8165 8166 8167 8168 8169 8170 8171 8172 8173 8174 8175 8176 8177 8178 8179 8180 8181 8182 8183 8184 8185 8186 8187 8188 8189 8190 8191 8192 8193 8194 8195 8196 8197
  }

  DBUG_PRINT("info",("looking for the following record"));
  DBUG_DUMP("record[0]", table->record[0], table->s->reclength);

  if ((table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
      table->s->primary_key < MAX_KEY)
  {
    /*
      Use a more efficient method to fetch the record given by
      table->record[0] if the engine allows it.  We first compute a
      row reference using the position() member function (it will be
      stored in table->file->ref) and the use rnd_pos() to position
      the "cursor" (i.e., record[0] in this case) at the correct row.

      TODO: Add a check that the correct record has been fetched by
      comparing with the original record. Take into account that the
      record on the master and slave can be of different
      length. Something along these lines should work:

      ADD>>>  store_record(table,record[1]);
              int error= table->file->ha_rnd_pos(table->record[0],
              table->file->ref);
      ADD>>>  DBUG_ASSERT(memcmp(table->record[1], table->record[0],
                                 table->s->reclength) == 0);

    */
    int error;
    DBUG_PRINT("info",("locating record using primary key (position)"));

    error= table->file->ha_rnd_pos_by_record(table->record[0]);
    if (unlikely(error))
    {
      DBUG_PRINT("info",("rnd_pos returns error %d",error));
      if (error == HA_ERR_KEY_NOT_FOUND)
        error= row_not_found_error(rgi);
      table->file->print_error(error, MYF(0));
    }
    DBUG_RETURN(error);
  }

  // We can't use position() - try other methods.
  
  /* 
    We need to retrieve all fields
    TODO: Move this out from this function to main loop 
   */
  table->use_all_columns();

  /*
    Save copy of the record in table->record[1]. It might be needed 
    later if linear search is used to find exact match.
   */ 
  store_record(table,record[1]);    

  if (m_key_info)
  {
    DBUG_PRINT("info",("locating record using key #%u [%s] (index_read)",
                       m_key_nr, m_key_info->name.str));
    /* We use this to test that the correct key is used in test cases. */
    DBUG_EXECUTE_IF("slave_crash_if_wrong_index",
                    if(0 != strcmp(m_key_info->name.str,"expected_key")) abort(););

    /* The key is active: search the table using the index */
    if (!table->file->inited &&
        (error= table->file->ha_index_init(m_key_nr, FALSE)))
    {
      DBUG_PRINT("info",("ha_index_init returns error %d",error));
      table->file->print_error(error, MYF(0));
      goto end;
    }

    /* Fill key data for the row */

    DBUG_ASSERT(m_key);
    key_copy(m_key, table->record[0], m_key_info, 0);

    /*
      Don't print debug messages when running valgrind since they can
      trigger false warnings.
     */
#ifndef HAVE_valgrind
    DBUG_DUMP("key data", m_key, m_key_info->key_length);
#endif

    /*
      We need to set the null bytes to ensure that the filler bit are
      all set when returning.  There are storage engines that just set
      the necessary bits on the bytes and don't set the filler bits
      correctly.
    */
    if (table->s->null_bytes > 0)
      table->record[0][table->s->null_bytes - 1]|=
        256U - (1U << table->s->last_null_bit_pos);

    if (unlikely((error= table->file->ha_index_read_map(table->record[0],
                                                        m_key,
                                                        HA_WHOLE_KEY,
                                                        HA_READ_KEY_EXACT))))
    {
      DBUG_PRINT("info",("no record matching the key found in the table"));
      if (error == HA_ERR_KEY_NOT_FOUND)
        error= row_not_found_error(rgi);
      table->file->print_error(error, MYF(0));
      table->file->ha_index_end();
      goto end;
    }

  /*
    Don't print debug messages when running valgrind since they can
    trigger false warnings.
   */
#ifndef HAVE_valgrind
    DBUG_PRINT("info",("found first matching record")); 
    DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
#endif
    /*
      Below is a minor "optimization".  If the key (i.e., key number
      0) has the HA_NOSAME flag set, we know that we have found the
      correct record (since there can be no duplicates); otherwise, we
      have to compare the record with the one found to see if it is
      the correct one.

      CAVEAT! This behaviour is essential for the replication of,
      e.g., the mysql.proc table since the correct record *shall* be
      found using the primary key *only*.  There shall be no
      comparison of non-PK columns to decide if the correct record is
      found.  I can see no scenario where it would be incorrect to
      chose the row to change only using a PK or an UNNI.
    */
    if (table->key_info->flags & HA_NOSAME)
    {
      /* Unique does not have non nullable part */
      if (!(table->key_info->flags & (HA_NULL_PART_KEY)))
      {
        error= 0;
        goto end;
      }
      else
      {
        KEY *keyinfo= table->key_info;
        /*
          Unique has nullable part. We need to check if there is any
          field in the BI image that is null and part of UNNI.
        */
        bool null_found= FALSE;
        for (uint i=0; i < keyinfo->user_defined_key_parts && !null_found; i++)
        {
          uint fieldnr= keyinfo->key_part[i].fieldnr - 1;
          Field **f= table->field+fieldnr;
          null_found= (*f)->is_null();
        }

        if (!null_found)
        {
          error= 0;
          goto end;
        }

        /* else fall through to index scan */
      }
    }

    is_index_scan=true;

    /*
      In case key is not unique, we still have to iterate over records found
      and find the one which is identical to the row given. A copy of the 
      record we are looking for is stored in record[1].
     */ 
    DBUG_PRINT("info",("non-unique index, scanning it to find matching record")); 
    /* We use this to test that the correct key is used in test cases. */
    DBUG_EXECUTE_IF("slave_crash_if_index_scan", abort(););

Marko Mäkelä's avatar
Marko Mäkelä committed
8198
    while (record_compare(table, m_vers_from_plain))
8199 8200 8201 8202
    {
      while ((error= table->file->ha_index_next(table->record[0])))
      {
        DBUG_PRINT("info",("no record matching the given row found"));
8203 8204
        if (error == HA_ERR_END_OF_FILE)
          error= end_of_file_error(rgi);
8205 8206 8207 8208 8209 8210 8211 8212 8213 8214 8215 8216 8217 8218 8219 8220 8221 8222 8223 8224 8225 8226 8227 8228 8229 8230 8231 8232 8233 8234 8235 8236 8237 8238 8239 8240
        table->file->print_error(error, MYF(0));
        table->file->ha_index_end();
        goto end;
      }
    }
  }
  else
  {
    DBUG_PRINT("info",("locating record using table scan (rnd_next)"));
    /* We use this to test that the correct key is used in test cases. */
    DBUG_EXECUTE_IF("slave_crash_if_table_scan", abort(););

    /* We don't have a key: search the table using rnd_next() */
    if (unlikely((error= table->file->ha_rnd_init_with_error(1))))
    {
      DBUG_PRINT("info",("error initializing table scan"
                         " (ha_rnd_init returns %d)",error));
      goto end;
    }

    is_table_scan= true;

    /* Continue until we find the right record or have made a full loop */
    do
    {
      error= table->file->ha_rnd_next(table->record[0]);

      if (unlikely(error))
        DBUG_PRINT("info", ("error: %s", HA_ERR(error)));
      switch (error) {

      case 0:
        DBUG_DUMP("record found", table->record[0], table->s->reclength);
        break;

      case HA_ERR_END_OF_FILE:
8241
        error= end_of_file_error(rgi);
8242 8243 8244 8245 8246 8247 8248 8249 8250 8251 8252 8253
        DBUG_PRINT("info", ("Record not found"));
        table->file->ha_rnd_end();
        goto end;

      default:
        DBUG_PRINT("info", ("Failed to get next record"
                            " (rnd_next returns %d)",error));
        table->file->print_error(error, MYF(0));
        table->file->ha_rnd_end();
        goto end;
      }
    }
Marko Mäkelä's avatar
Marko Mäkelä committed
8254
    while (record_compare(table, m_vers_from_plain));
8255 8256 8257 8258 8259 8260 8261 8262 8263 8264 8265 8266 8267 8268 8269 8270 8271 8272 8273 8274 8275 8276 8277
    
    /* 
      Note: above record_compare will take into accout all record fields 
      which might be incorrect in case a partial row was given in the event
     */

    DBUG_ASSERT(error == HA_ERR_END_OF_FILE || error == 0);
  }

end:
  if (is_table_scan || is_index_scan)
    issue_long_find_row_warning(get_general_type_code(), m_table->alias.c_ptr(), 
                                is_index_scan, rgi);
  DBUG_RETURN(error);
}

#endif

/*
  Constructor used to build an event for writing to the binary log.
 */

Delete_rows_log_event::Delete_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
8278 8279
                                             ulonglong tid,
                                             bool is_transactional)
8280 8281 8282 8283 8284 8285 8286
  : Rows_log_event(thd_arg, tbl_arg, tid, tbl_arg->read_set, is_transactional,
                   DELETE_ROWS_EVENT_V1)
{
}

Delete_rows_compressed_log_event::Delete_rows_compressed_log_event(
                                           THD *thd_arg, TABLE *tbl_arg,
8287
                                           ulonglong tid_arg,
8288 8289 8290 8291 8292 8293 8294 8295 8296 8297 8298 8299 8300 8301 8302 8303 8304 8305 8306 8307 8308 8309 8310 8311 8312 8313 8314 8315 8316 8317 8318
                                           bool is_transactional)
  : Delete_rows_log_event(thd_arg, tbl_arg, tid_arg, is_transactional)
{
  m_type= DELETE_ROWS_COMPRESSED_EVENT_V1;
}

bool Delete_rows_compressed_log_event::write()
{
  return Rows_log_event::write_compressed();    
}


#if defined(HAVE_REPLICATION)

int 
Delete_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
  /*
    Increment the global status delete count variable
   */
  if (get_flags(STMT_END_F))
    status_var_increment(thd->status_var.com_stat[SQLCOM_DELETE]);

  if ((m_table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
      m_table->s->primary_key < MAX_KEY)
  {
    /*
      We don't need to allocate any memory for m_key since it is not used.
    */
    return 0;
  }
8319
  if (do_invoke_trigger())
8320 8321 8322 8323 8324 8325 8326 8327 8328 8329 8330 8331 8332 8333 8334 8335 8336 8337 8338 8339 8340
    m_table->prepare_triggers_for_delete_stmt_or_event();

  return find_key();
}

int 
Delete_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const, 
                                               int error)
{
  m_table->file->ha_index_or_rnd_end();
  my_free(m_key);
  m_key= NULL;
  m_key_info= NULL;

  return error;
}

int Delete_rows_log_event::do_exec_row(rpl_group_info *rgi)
{
  int error;
  const char *tmp= thd->get_proc_info();
8341
  char *message, msg[128];
Marko Mäkelä's avatar
Marko Mäkelä committed
8342 8343 8344 8345 8346 8347
  const LEX_CSTRING &table_name= m_table->s->table_name;
  const char quote_char=
    get_quote_char_for_identifier(thd, table_name.str, table_name.length);
  my_snprintf(msg, sizeof msg,
              "Delete_rows_log_event::find_row() on table %c%.*s%c",
              quote_char, int(table_name.length), table_name.str, quote_char);
8348
  message= msg;
8349
  const bool invoke_triggers= (m_table->triggers && do_invoke_trigger());
8350 8351 8352 8353
  DBUG_ASSERT(m_table != NULL);

#ifdef WSREP_PROC_INFO
  my_snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
8354 8355 8356
              "Delete_rows_log_event::find_row(%lld) on table %c%.*s%c",
              (long long) wsrep_thd_trx_seqno(thd), quote_char,
              int(table_name.length), table_name.str,
8357
              quote_char);
8358 8359 8360 8361 8362
  message= thd->wsrep_info;
#endif /* WSREP_PROC_INFO */

  thd_proc_info(thd, message);
  if (likely(!(error= find_row(rgi))))
Marko Mäkelä's avatar
Marko Mäkelä committed
8363
  {
8364 8365 8366
    /*
      Delete the record found, located in record[0]
    */
Marko Mäkelä's avatar
Marko Mäkelä committed
8367 8368 8369 8370
    my_snprintf(msg, sizeof msg,
                "Delete_rows_log_event::ha_delete_row() on table %c%.*s%c",
                quote_char, int(table_name.length), table_name.str,
                quote_char);
8371
    message= msg;
8372 8373
#ifdef WSREP_PROC_INFO
    snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
8374 8375 8376
             "Delete_rows_log_event::ha_delete_row(%lld) on table %c%.*s%c",
             (long long) wsrep_thd_trx_seqno(thd), quote_char,
             int(table_name.length), table_name.str, quote_char);
8377 8378 8379 8380 8381 8382 8383 8384 8385 8386 8387 8388 8389 8390 8391 8392 8393 8394 8395 8396 8397 8398 8399 8400 8401 8402 8403 8404 8405 8406 8407 8408 8409 8410 8411 8412 8413 8414 8415 8416 8417 8418 8419 8420 8421 8422 8423 8424 8425 8426 8427
    message= thd->wsrep_info;
#endif
    thd_proc_info(thd, message);

    if (invoke_triggers &&
        unlikely(process_triggers(TRG_EVENT_DELETE, TRG_ACTION_BEFORE, FALSE)))
      error= HA_ERR_GENERIC; // in case if error is not set yet
    if (likely(!error))
    {
      m_table->mark_columns_per_binlog_row_image();
      if (m_vers_from_plain && m_table->versioned(VERS_TIMESTAMP))
      {
        Field *end= m_table->vers_end_field();
        bitmap_set_bit(m_table->write_set, end->field_index);
        store_record(m_table, record[1]);
        end->set_time();
        error= m_table->file->ha_update_row(m_table->record[1],
                                            m_table->record[0]);
      }
      else
      {
        error= m_table->file->ha_delete_row(m_table->record[0]);
      }
      m_table->default_column_bitmaps();
    }
    if (invoke_triggers && likely(!error) &&
        unlikely(process_triggers(TRG_EVENT_DELETE, TRG_ACTION_AFTER, FALSE)))
      error= HA_ERR_GENERIC; // in case if error is not set yet
    m_table->file->ha_index_or_rnd_end();
  }
  thd_proc_info(thd, tmp);
  return error;
}

#endif /* defined(HAVE_REPLICATION) */

#if defined(HAVE_REPLICATION)
uint8 Delete_rows_log_event::get_trg_event_map()
{
  return trg2bit(TRG_EVENT_DELETE);
}
#endif

/**************************************************************************
	Update_rows_log_event member functions
**************************************************************************/

/*
  Constructor used to build an event for writing to the binary log.
 */
Update_rows_log_event::Update_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
8428
                                             ulonglong tid,
8429 8430 8431 8432 8433 8434 8435
                                             bool is_transactional)
: Rows_log_event(thd_arg, tbl_arg, tid, tbl_arg->read_set, is_transactional,
                 UPDATE_ROWS_EVENT_V1)
{
  init(tbl_arg->rpl_write_set);
}

8436 8437 8438
Update_rows_compressed_log_event::
Update_rows_compressed_log_event(THD *thd_arg, TABLE *tbl_arg,
                                 ulonglong tid, bool is_transactional)
8439 8440 8441 8442 8443 8444 8445 8446 8447 8448 8449 8450 8451 8452 8453 8454 8455 8456 8457 8458
: Update_rows_log_event(thd_arg, tbl_arg, tid, is_transactional)
{
  m_type = UPDATE_ROWS_COMPRESSED_EVENT_V1;
}

bool Update_rows_compressed_log_event::write()
{
  return Rows_log_event::write_compressed();
}

void Update_rows_log_event::init(MY_BITMAP const *cols)
{
  /* if my_bitmap_init fails, caught in is_valid() */
  if (likely(!my_bitmap_init(&m_cols_ai,
                          m_width <= sizeof(m_bitbuf_ai)*8 ? m_bitbuf_ai : NULL,
                          m_width,
                          false)))
  {
    /* Cols can be zero if this is a dummy binrows event */
    if (likely(cols != NULL))
8459
      bitmap_copy(&m_cols_ai, cols);
8460 8461 8462 8463 8464 8465 8466 8467 8468 8469 8470 8471 8472 8473 8474 8475 8476 8477 8478
  }
}


#if defined(HAVE_REPLICATION)

int 
Update_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
  /*
    Increment the global status update count variable
  */
  if (get_flags(STMT_END_F))
    status_var_increment(thd->status_var.com_stat[SQLCOM_UPDATE]);

  int err;
  if ((err= find_key()))
    return err;

8479
  if (do_invoke_trigger())
8480 8481 8482 8483 8484 8485 8486 8487 8488 8489 8490 8491 8492 8493 8494 8495 8496 8497
    m_table->prepare_triggers_for_update_stmt_or_event();

  return 0;
}

int 
Update_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const, 
                                               int error)
{
  /*error= ToDo:find out what this should really be, this triggers close_scan in nbd, returning error?*/
  m_table->file->ha_index_or_rnd_end();
  my_free(m_key); // Free for multi_malloc
  m_key= NULL;
  m_key_info= NULL;

  return error;
}

8498
int
8499 8500
Update_rows_log_event::do_exec_row(rpl_group_info *rgi)
{
8501
  const bool invoke_triggers= (m_table->triggers && do_invoke_trigger());
8502 8503
  const char *tmp= thd->get_proc_info();
  DBUG_ASSERT(m_table != NULL);
8504
  char *message, msg[128];
Marko Mäkelä's avatar
Marko Mäkelä committed
8505 8506 8507 8508 8509 8510
  const LEX_CSTRING &table_name= m_table->s->table_name;
  const char quote_char=
    get_quote_char_for_identifier(thd, table_name.str, table_name.length);
  my_snprintf(msg, sizeof msg,
              "Update_rows_log_event::find_row() on table %c%.*s%c",
              quote_char, int(table_name.length), table_name.str, quote_char);
8511
  message= msg;
8512 8513 8514

#ifdef WSREP_PROC_INFO
  my_snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
8515 8516 8517
              "Update_rows_log_event::find_row(%lld) on table %c%.*s%c",
              (long long) wsrep_thd_trx_seqno(thd), quote_char,
              int(table_name.length), table_name.str,
8518
              quote_char);
8519 8520 8521 8522
  message= thd->wsrep_info;
#endif /* WSREP_PROC_INFO */

  thd_proc_info(thd, message);
Marko Mäkelä's avatar
Marko Mäkelä committed
8523 8524 8525 8526 8527 8528 8529
  // Temporary fix to find out why it fails [/Matz]
  memcpy(m_table->read_set->bitmap, m_cols.bitmap, (m_table->read_set->n_bits + 7) / 8);
  memcpy(m_table->write_set->bitmap, m_cols_ai.bitmap, (m_table->write_set->n_bits + 7) / 8);

  m_table->mark_columns_per_binlog_row_image();

  int error= find_row(rgi);
8530 8531 8532 8533 8534 8535 8536 8537 8538 8539 8540 8541
  if (unlikely(error))
  {
    /*
      We need to read the second image in the event of error to be
      able to skip to the next pair of updates
    */
    if ((m_curr_row= m_curr_row_end))
      unpack_current_row(rgi, &m_cols_ai);
    thd_proc_info(thd, tmp);
    return error;
  }

8542 8543 8544 8545 8546
  const bool history_change= m_table->versioned() ?
    !m_table->vers_end_field()->is_max() : false;
  TABLE_LIST *tl= m_table->pos_in_table_list;
  uint8 trg_event_map_save= tl->trg_event_map;

8547 8548 8549 8550 8551 8552 8553 8554 8555 8556 8557 8558 8559 8560
  /*
    This is the situation after locating BI:

    ===|=== before image ====|=== after image ===|===
       ^                     ^
       m_curr_row            m_curr_row_end

    BI found in the table is stored in record[0]. We copy it to record[1]
    and unpack AI to record[0].
   */

  store_record(m_table,record[1]);

  m_curr_row= m_curr_row_end;
Marko Mäkelä's avatar
Marko Mäkelä committed
8561 8562 8563
  my_snprintf(msg, sizeof msg,
              "Update_rows_log_event::unpack_current_row() on table %c%.*s%c",
              quote_char, int(table_name.length), table_name.str, quote_char);
8564
  message= msg;
8565 8566
#ifdef WSREP_PROC_INFO
  my_snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
8567 8568 8569
              "Update_rows_log_event::unpack_current_row(%lld) on table %c%.*s%c",
              (long long) wsrep_thd_trx_seqno(thd), quote_char,
              int(table_name.length), table_name.str, quote_char);
8570 8571 8572 8573 8574 8575 8576
  message= thd->wsrep_info;
#endif /* WSREP_PROC_INFO */

  /* this also updates m_curr_row_end */
  thd_proc_info(thd, message);
  if (unlikely((error= unpack_current_row(rgi, &m_cols_ai))))
    goto err;
Marko Mäkelä's avatar
Marko Mäkelä committed
8577 8578
  if (m_table->s->long_unique_table)
    m_table->update_virtual_fields(m_table->file, VCOL_UPDATE_FOR_WRITE);
8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592 8593

  /*
    Now we have the right row to update.  The old row (the one we're
    looking for) is in record[1] and the new row is in record[0].
  */
#ifndef HAVE_valgrind
  /*
    Don't print debug messages when running valgrind since they can
    trigger false warnings.
   */
  DBUG_PRINT("info",("Updating row in table"));
  DBUG_DUMP("old record", m_table->record[1], m_table->s->reclength);
  DBUG_DUMP("new values", m_table->record[0], m_table->s->reclength);
#endif

Marko Mäkelä's avatar
Marko Mäkelä committed
8594 8595 8596
  my_snprintf(msg, sizeof msg,
              "Update_rows_log_event::ha_update_row() on table %c%.*s%c",
              quote_char, int(table_name.length), table_name.str, quote_char);
8597
  message= msg;
8598 8599
#ifdef WSREP_PROC_INFO
  my_snprintf(thd->wsrep_info, sizeof(thd->wsrep_info) - 1,
Marko Mäkelä's avatar
Marko Mäkelä committed
8600 8601 8602
              "Update_rows_log_event::ha_update_row(%lld) on table %c%.*s%c",
              (long long) wsrep_thd_trx_seqno(thd), quote_char,
              int(table_name.length), table_name.str, quote_char);
8603 8604 8605 8606 8607 8608 8609 8610 8611 8612 8613
  message= thd->wsrep_info;
#endif /* WSREP_PROC_INFO */

  thd_proc_info(thd, message);
  if (invoke_triggers &&
      unlikely(process_triggers(TRG_EVENT_UPDATE, TRG_ACTION_BEFORE, TRUE)))
  {
    error= HA_ERR_GENERIC; // in case if error is not set yet
    goto err;
  }

8614 8615 8616 8617 8618 8619 8620 8621 8622
  if (m_table->versioned())
  {
    if (m_vers_from_plain && m_table->versioned(VERS_TIMESTAMP))
      m_table->vers_update_fields();
    if (!history_change && !m_table->vers_end_field()->is_max())
    {
      tl->trg_event_map|= trg2bit(TRG_EVENT_DELETE);
    }
  }
8623
  error= m_table->file->ha_update_row(m_table->record[1], m_table->record[0]);
8624
  tl->trg_event_map= trg_event_map_save;
8625 8626 8627 8628 8629 8630 8631 8632 8633 8634 8635 8636 8637 8638 8639 8640
  if (unlikely(error == HA_ERR_RECORD_IS_THE_SAME))
    error= 0;
  if (m_vers_from_plain && m_table->versioned(VERS_TIMESTAMP))
  {
    store_record(m_table, record[2]);
    error= vers_insert_history_row(m_table);
    restore_record(m_table, record[2]);
  }
  m_table->default_column_bitmaps();

  if (invoke_triggers && likely(!error) &&
      unlikely(process_triggers(TRG_EVENT_UPDATE, TRG_ACTION_AFTER, TRUE)))
    error= HA_ERR_GENERIC; // in case if error is not set yet


err:
8641
  thd_proc_info(thd, tmp);
8642 8643 8644 8645 8646 8647 8648 8649 8650 8651 8652 8653 8654 8655 8656 8657 8658 8659 8660 8661 8662 8663 8664 8665 8666 8667 8668 8669 8670 8671 8672 8673 8674 8675 8676 8677 8678 8679 8680 8681 8682
  m_table->file->ha_index_or_rnd_end();
  return error;
}

#endif /* defined(HAVE_REPLICATION) */


#if defined(HAVE_REPLICATION)
uint8 Update_rows_log_event::get_trg_event_map()
{
  return trg2bit(TRG_EVENT_UPDATE);
}
#endif


void Incident_log_event::pack_info(Protocol *protocol)
{
  char buf[256];
  size_t bytes;
  if (m_message.length > 0)
    bytes= my_snprintf(buf, sizeof(buf), "#%d (%s)",
                       m_incident, description());
  else
    bytes= my_snprintf(buf, sizeof(buf), "#%d (%s): %s",
                       m_incident, description(), m_message.str);
  protocol->store(buf, bytes, &my_charset_bin);
}


#if defined(WITH_WSREP)
/*
  read the first event from (*buf). The size of the (*buf) is (*buf_len).
  At the end (*buf) is shitfed to point to the following event or NULL and
  (*buf_len) will be changed to account just being read bytes of the 1st event.
*/
#define WSREP_MAX_ALLOWED_PACKET 1024*1024*1024 // current protocol max

Log_event* wsrep_read_log_event(
  char **arg_buf, size_t *arg_buf_len,
  const Format_description_log_event *description_event)
{
8683
  uchar *head= (uchar*) (*arg_buf);
8684 8685 8686 8687 8688 8689 8690 8691 8692 8693 8694
  uint data_len = uint4korr(head + EVENT_LEN_OFFSET);
  const char *error= 0;
  Log_event *res=  0;
  DBUG_ENTER("wsrep_read_log_event");

  if (data_len > WSREP_MAX_ALLOWED_PACKET)
  {
    error = "Event too big";
    goto err;
  }

8695 8696
  res= Log_event::read_log_event(head, data_len, &error, description_event,
                                 false);
8697 8698 8699 8700 8701 8702

err:
  if (!res)
  {
    DBUG_ASSERT(error != 0);
    sql_print_error("Error in Log_event::read_log_event(): "
8703 8704
                    "'%s', data_len: %u, event_type: %d",
		    error, data_len, (int) head[EVENT_TYPE_OFFSET]);
8705 8706 8707 8708 8709 8710 8711 8712 8713
  }
  (*arg_buf)+= data_len;
  (*arg_buf_len)-= data_len;
  DBUG_RETURN(res);
}
#endif


#if defined(HAVE_REPLICATION)
8714
int Incident_log_event::do_apply_event(rpl_group_info *rgi)
8715 8716 8717 8718 8719 8720 8721 8722 8723 8724 8725 8726 8727 8728 8729 8730 8731 8732 8733 8734 8735 8736 8737 8738 8739 8740 8741 8742 8743 8744 8745 8746 8747 8748 8749 8750 8751 8752 8753 8754 8755 8756 8757 8758 8759 8760 8761 8762 8763 8764 8765 8766
{
  Relay_log_info const *rli= rgi->rli;
  DBUG_ENTER("Incident_log_event::do_apply_event");

  if (ignored_error_code(ER_SLAVE_INCIDENT))
  {
    DBUG_PRINT("info", ("Ignoring Incident"));
    DBUG_RETURN(0);
  }

  rli->report(ERROR_LEVEL, ER_SLAVE_INCIDENT, NULL,
              ER_THD(rgi->thd, ER_SLAVE_INCIDENT),
              description(),
              m_message.length > 0 ? m_message.str : "<none>");
  DBUG_RETURN(1);
}
#endif


bool
Incident_log_event::write_data_header()
{
  DBUG_ENTER("Incident_log_event::write_data_header");
  DBUG_PRINT("enter", ("m_incident: %d", m_incident));
  uchar buf[sizeof(int16)];
  int2store(buf, (int16) m_incident);
  DBUG_RETURN(write_data(buf, sizeof(buf)));
}

bool
Incident_log_event::write_data_body()
{
  uchar tmp[1];
  DBUG_ENTER("Incident_log_event::write_data_body");
  tmp[0]= (uchar) m_message.length;
  DBUG_RETURN(write_data(tmp, sizeof(tmp)) ||
              write_data(m_message.str, m_message.length));
}


/* Pack info for its unrecognized ignorable event */
void Ignorable_log_event::pack_info(Protocol *protocol)
{
  char buf[256];
  size_t bytes;
  bytes= my_snprintf(buf, sizeof(buf), "# Ignorable event type %d (%s)",
                     number, description);
  protocol->store(buf, bytes, &my_charset_bin);
}


#if defined(HAVE_REPLICATION)
8767
Heartbeat_log_event::Heartbeat_log_event(const uchar *buf, uint event_len,
8768 8769 8770 8771
                    const Format_description_log_event* description_event)
  :Log_event(buf, description_event)
{
  uint8 header_size= description_event->common_header_len;
8772 8773 8774 8775 8776 8777 8778 8779 8780 8781 8782
  if (log_pos == 0)
  {
    log_pos= uint8korr(buf + header_size);
    log_ident= buf + header_size + HB_SUB_HEADER_LEN;
    ident_len= event_len - (header_size + HB_SUB_HEADER_LEN);
  }
  else
  {
    log_ident= buf + header_size;
    ident_len = event_len - header_size;
  }
8783 8784 8785 8786 8787 8788 8789 8790 8791 8792 8793 8794 8795 8796
}
#endif


/**
   Check if we should write event to the relay log

   This is used to skip events that is only supported by MySQL

   Return:
   0 ok
   1 Don't write event
*/

8797
bool event_that_should_be_ignored(const uchar *buf)
8798
{
8799
  uint event_type= buf[EVENT_TYPE_OFFSET];
8800 8801 8802 8803 8804 8805 8806 8807 8808
  if (event_type == GTID_LOG_EVENT ||
      event_type == ANONYMOUS_GTID_LOG_EVENT ||
      event_type == PREVIOUS_GTIDS_LOG_EVENT ||
      event_type == TRANSACTION_CONTEXT_EVENT ||
      event_type == VIEW_CHANGE_EVENT ||
      (uint2korr(buf + FLAGS_OFFSET) & LOG_EVENT_IGNORABLE_F))
    return 1;
  return 0;
}