mariadb/storage/innobase/trx/trx0sys.c
unknown 76de7d788c Apply the following InnoDB snapshots:
innodb-5.1-ss1318
innodb-5.1-ss1330
innodb-5.1-ss1332
innodb-5.1-ss1340

Fixes:
- Bug #21409: Incorrect result returned when in READ-COMMITTED with query_cache ON
  At low transaction isolation levels we let each consistent read set
  its own snapshot.

- Bug #23666: strange Innodb_row_lock_time_% values in show status; also millisecs wrong
  On Windows ut_usectime returns secs and usecs relative to the UNIX
  epoch (which is Jan, 1 1970).

- Bug #25494: LATEST DEADLOCK INFORMATION is not always cleared
  lock_deadlock_recursive(): When the search depth or length is exceeded,
  rewind lock_latest_err_file and display the two transactions at the
  point of aborting the search.

- Bug #25927: Foreign key with ON DELETE SET NULL on NOT NULL can crash server
  Prevent ALTER TABLE ... MODIFY ... NOT NULL on columns for which
  there is a foreign key constraint ON ... SET NULL.

- Bug #26835: Repeatable corruption of utf8-enabled tables inside InnoDB
  The bug could be reproduced as follows:

  Define a table so that the first column of the clustered index is
  a VARCHAR or a UTF-8 CHAR in a collation where sequences of bytes
  of differing length are considered equivalent.

  Insert and delete a record.  Before the delete-marked record is
  purged, insert another record whose first column is of different
  length but equivalent to the first record.  Under certain conditions,
  the insertion can be incorrectly performed as update-in-place.

  Likewise, an operation that could be done as update-in-place can
  unnecessarily be performed as delete and insert, but that would not
  cause corruption but merely degraded performance.


mysql-test/r/innodb.result:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1284:
  Merge changes from MySQL AB:
  
  ChangeSet
    2007/01/24 14:49:36+04:00 holyfoot@mysql.com 
    bug 22682 Test fails --without-geometry
    geometry dependent parts moved to proper .test files
  
  mysql-test/r/innodb.result
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +0 -2
    result fixed
  
  mysql-test/r/innodb_gis.result
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +2 -0
    result fixed
  
  mysql-test/t/innodb.test
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +0 -6
    HAVE_GEOMETRY dependent part moved to innodb_gis.test
  
  mysql-test/t/innodb_gis.test
    2007/01/24 14:49:35+04:00 holyfoot@mysql.com +6 -0
    HAVE_GEOMETRY dependent part moved here from innodb.test
  
  
  Revision r1186:
  dict_load_foreign(): Use a local variable instead of the 10-bit field
  foreign->n_fields in order to preserve ON UPDATE CASCADE and
  ON DELETE CASCADE flags.  For some reason, gcc does not warn about
  shifting a 10-bit field to right by 24 bits.  (Bug 24741)
  
  This bug was introduced while reducing the memory footprint of the
  InnoDB data dictionary (Bug 20877).
  
  innodb.test, innodb.result: Add a test case.
  
  
  Revision r1318:
  Add a test case for r1316 (Bug #25927).
  
  
  Revision r1340:
  innodb.test, innodb.result: Add test case for Bug #26835.
  The bug could be reproduced as follows:
  
  Define a table so that the first column of the clustered index is
  a VARCHAR or a UTF-8 CHAR in a collation where sequences of bytes
  of differing length are considered equivalent.
  
  Insert and delete a record.  Before the delete-marked record is
  purged, insert another record whose first column is of different
  length but equivalent to the first record.  Under certain conditions,
  the insertion can be incorrectly performed as update-in-place.
  
  Likewise, an operation that could be done as update-in-place can
  unnecessarily be performed as delete and insert, but that would not
  cause corruption but merely degraded performance.
mysql-test/t/innodb.test:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1284:
  Merge changes from MySQL AB:
  
  ChangeSet
    2007/01/24 14:49:36+04:00 holyfoot@mysql.com 
    bug 22682 Test fails --without-geometry
    geometry dependent parts moved to proper .test files
  
  mysql-test/r/innodb.result
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +0 -2
    result fixed
  
  mysql-test/r/innodb_gis.result
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +2 -0
    result fixed
  
  mysql-test/t/innodb.test
    2007/01/24 14:49:34+04:00 holyfoot@mysql.com +0 -6
    HAVE_GEOMETRY dependent part moved to innodb_gis.test
  
  mysql-test/t/innodb_gis.test
    2007/01/24 14:49:35+04:00 holyfoot@mysql.com +6 -0
    HAVE_GEOMETRY dependent part moved here from innodb.test
  
  
  Revision r1283:
  Merge changes from MySQL AB:
  
  ChangeSet
    2007/01/22 18:42:52+02:00 monty@mysql.com 
    Give warnings for unused objects
    Changed error message to be compatible with old error file
    Added new error message for new DUP_ENTRY syntax
  
  mysql-test/t/innodb.test
    2007/01/22 18:42:49+02:00 monty@mysql.com +14 -14
    Changed to use new error message
  
  
  Revision r1186:
  dict_load_foreign(): Use a local variable instead of the 10-bit field
  foreign->n_fields in order to preserve ON UPDATE CASCADE and
  ON DELETE CASCADE flags.  For some reason, gcc does not warn about
  shifting a 10-bit field to right by 24 bits.  (Bug 24741)
  
  This bug was introduced while reducing the memory footprint of the
  InnoDB data dictionary (Bug 20877).
  
  innodb.test, innodb.result: Add a test case.
  
  
  Revision r1318:
  Add a test case for r1316 (Bug #25927).
  
  
  Revision r1329:
  Merge changes from MySQL AB to mysql-test directives.
  The results are not affected.
  
  
  Revision r1340:
  innodb.test, innodb.result: Add test case for Bug #26835.
  The bug could be reproduced as follows:
  
  Define a table so that the first column of the clustered index is
  a VARCHAR or a UTF-8 CHAR in a collation where sequences of bytes
  of differing length are considered equivalent.
  
  Insert and delete a record.  Before the delete-marked record is
  purged, insert another record whose first column is of different
  length but equivalent to the first record.  Under certain conditions,
  the insertion can be incorrectly performed as update-in-place.
  
  Likewise, an operation that could be done as update-in-place can
  unnecessarily be performed as delete and insert, but that would not
  cause corruption but merely degraded performance.
storage/innobase/buf/buf0buf.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/buf/buf0flu.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/buf/buf0lru.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/dict/dict0boot.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/dict/dict0crea.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/dict/dict0dict.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1316:
  Prevent ALTER TABLE ... MODIFY ... NOT NULL on columns for which
  there is a foreign key constraint ON ... SET NULL.  (Bug #25927)
  
  dict_foreign_find_index(): Add paramettter check_null.
  
  dict_foreign_add_to_cache(): Do not allow ON DELETE SET NULL
  or ON UPDATE SET NULL if any of the referencing columns are declared NOT NULL.
  
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/dict/dict0load.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1186:
  dict_load_foreign(): Use a local variable instead of the 10-bit field
  foreign->n_fields in order to preserve ON UPDATE CASCADE and
  ON DELETE CASCADE flags.  For some reason, gcc does not warn about
  shifting a 10-bit field to right by 24 bits.  (Bug 24741)
  
  This bug was introduced while reducing the memory footprint of the
  InnoDB data dictionary (Bug 20877).
  
  innodb.test, innodb.result: Add a test case.
  
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/fil/fil0fil.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/fsp/fsp0fsp.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/ha/ha0ha.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/handler/ha_innodb.cc:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1204:
  Change this in ha_innobase:
  
   void*           innobase_prebuilt;
  
  to this:
  
   row_prebuilt_t* prebuilt;
  
  by introducing the typedef in ha_innodb.h, and remove all the now needless
  local variables and casts in ha_innodb.cc.
  
  Revision r1298:
  ha_innodb.cc: Remove all references to thd->ha_data[hton->slot].
  
  thd_to_trx(thd, hton): Accessor for getting the InnoDB trx object
  of a MySQL thread object and an InnoDB handlerton.
  
  
  Revision r1292:
  Remove the declarations of some global functions in ha_innodb.h and declare
  them static in ha_innodb.cc.  These functions are invoked via function
  pointers in handlerton.
  
  
  Revision r1300:
  ha_innodb.cc: Replace thd->tablespace_op with thd_tablespace_op(thd).
  Plugins must treat class THD as an opaque type.
  
  
  Revision r1198:
  Merge a change from MySQL AB:
  
  ChangeSet@1.2372, 2006-12-31 02:29:11+01:00, kent@mysql.com +79 -0
    Many files:
      Removed "MySQL Finland AB & TCX DataKonsult AB" from copyright header
      Adjusted year(s) in copyright header 
      Added GPL copyright text
  
  
  Revision r1271:
  Merge changes from MySQL AB:
  
  Rename some FIELD_TYPE_ constants to MYSQL_TYPE_.
  
  Change the scope of a type cast of two dividends.
  
  
  Revision r1299:
  ha_innodb.cc: Replace thd->in_lock_tables with thd_in_lock_tables(thd).
  Plugins must treat class THD as an opaque type.
  
  
  Revision r1201:
  Apply patch from MySQL:
  
   ChangeSet@1.2353, 2006-12-19 16:57:51-07:00, tsmith@siva.hindu.god +13 -0
     Added innodb_rollback_on_timeout option to restore the 4.1 
     InnoDB timeout behavior (Bug 24200)
  
  Revision r1322:
  ha_innodb.cc: Remove the unused innobase_repl_ variables.
  
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
  
  
  Revision r1334:
  Fix for Bug# 21409. At low transaction isolation levels we let each 
  consistent read set its own snapshot
storage/innobase/handler/ha_innodb.h:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1204:
  Change this in ha_innobase:
  
   void*           innobase_prebuilt;
  
  to this:
  
   row_prebuilt_t* prebuilt;
  
  by introducing the typedef in ha_innodb.h, and remove all the now needless
  local variables and casts in ha_innodb.cc.
  
  Revision r1292:
  Remove the declarations of some global functions in ha_innodb.h and declare
  them static in ha_innodb.cc.  These functions are invoked via function
  pointers in handlerton.
  
  
  Revision r1198:
  Merge a change from MySQL AB:
  
  ChangeSet@1.2372, 2006-12-31 02:29:11+01:00, kent@mysql.com +79 -0
    Many files:
      Removed "MySQL Finland AB & TCX DataKonsult AB" from copyright header
      Adjusted year(s) in copyright header 
      Added GPL copyright text
  
  
  Revision r1201:
  Apply patch from MySQL:
  
   ChangeSet@1.2353, 2006-12-19 16:57:51-07:00, tsmith@siva.hindu.god +13 -0
     Added innodb_rollback_on_timeout option to restore the 4.1 
     InnoDB timeout behavior (Bug 24200)
storage/innobase/ibuf/ibuf0ibuf.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/buf0buf.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/buf0flu.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/dict0dict.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/ha0ha.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/lock0lock.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/log0log.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/mem0mem.h:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1241:
  Remove the unused function mem_strdupq().
storage/innobase/include/mem0mem.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1241:
  Remove the unused function mem_strdupq().
storage/innobase/include/rem0rec.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1338:
  rec_offs_nth_size(): Treat n==0 as a special case.  (Bug #26835)
storage/innobase/include/sync0rw.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/sync0sync.h:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1247:
  Rename mutex_enter_nowait to mutex_enter_nowait_func and add macro
  mutex_enter_nowait that supplies the default __FILE__ and __LINE__
  arguments. Adjust callers.
storage/innobase/include/sync0sync.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1294:
  Fixed inline asm code, it didn't work with GCC > ver 3.x.
  
  
  Revision r1244:
  Add ut_ad() debug assertions.
  
  UT_LIST_ADD_FIRST(), UT_LIST_ADD_LAST(), UT_LIST_INSERT_AFTER():
  Assert against some trivial cases of cyclic lists.
  
  mutex_enter_func(): Assert that the current thread is not holding the mutex.
  
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/trx0sys.ic:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/include/univ.i:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1285:
  Merge a change from MySQL AB:
  
  ChangeSet
    2006/10/26 15:41:47-04:00 iggy@amd64. 
    Post Merge Cleanup
  
  storage/innobase/include/univ.i
    2006/10/26 15:38:50-04:00 iggy@amd64. +9 -0
    Post Merge Cleanup
storage/innobase/include/ut0lst.h:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1244:
  Add ut_ad() debug assertions.
  
  UT_LIST_ADD_FIRST(), UT_LIST_ADD_LAST(), UT_LIST_INSERT_AFTER():
  Assert against some trivial cases of cyclic lists.
  
  mutex_enter_func(): Assert that the current thread is not holding the mutex.
storage/innobase/lock/lock0lock.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1330:
  lock_deadlock_recursive(): When the search depth or length is exceeded,
  rewind lock_latest_err_file and display the two transactions at the
  point of aborting the search.  (Bug #25494)
  
  
  Revision r1332:
  lock_deadlock_recursive(): When aborting the search, display a note
  regardless of start->undo_no.  Otherwise, aborted searches may show
  up as genuine deadlocks.  This mistake was made in r1330.
storage/innobase/log/log0log.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1247:
  Rename mutex_enter_nowait to mutex_enter_nowait_func and add macro
  mutex_enter_nowait that supplies the default __FILE__ and __LINE__
  arguments. Adjust callers.
storage/innobase/log/log0recv.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/mem/mem0pool.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/pars/pars0pars.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/que/que0que.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/read/read0read.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/row/row0mysql.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1201:
  Apply patch from MySQL:
  
   ChangeSet@1.2353, 2006-12-19 16:57:51-07:00, tsmith@siva.hindu.god +13 -0
     Added innodb_rollback_on_timeout option to restore the 4.1 
     InnoDB timeout behavior (Bug 24200)
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/row/row0vers.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/srv/srv0que.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/srv/srv0srv.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1262:
  Fix for Bug# 23666. On Windows ut_usectime returns secs 
  and usecs relative to the UNIX epoch (which is Jan, 1 1970).
  
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/sync/sync0rw.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1247:
  Rename mutex_enter_nowait to mutex_enter_nowait_func and add macro
  mutex_enter_nowait that supplies the default __FILE__ and __LINE__
  arguments. Adjust callers.
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/sync/sync0sync.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1247:
  Rename mutex_enter_nowait to mutex_enter_nowait_func and add macro
  mutex_enter_nowait that supplies the default __FILE__ and __LINE__
  arguments. Adjust callers.
storage/innobase/thr/thr0loc.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/trx/trx0purge.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/trx/trx0roll.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/trx/trx0rseg.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/trx/trx0sys.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/trx/trx0trx.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
  
  
  Revision r1324:
  Merge changes from MySQL AB:
  
  ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
    Fixed compiler warnings
    ...
    Fixed compiler warnings detected on windows64
storage/innobase/trx/trx0undo.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/usr/usr0sess.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1242:
  Merge r1239 from
  branches/zip: Make mutex_own() work with UNIV_DEBUG, without UNIV_SYNC_DEBUG.
storage/innobase/ut/ut0ut.c:
  Apply the following InnoDB snapshots:
  innodb-5.1-ss1318
  innodb-5.1-ss1330
  innodb-5.1-ss1332
  innodb-5.1-ss1340
  
  Revision r1262:
  Fix for Bug# 23666. On Windows ut_usectime returns secs 
  and usecs relative to the UNIX epoch (which is Jan, 1 1970).
2007-03-22 15:59:35 -06:00

986 lines
26 KiB
C

/******************************************************
Transaction system
(c) 1996 Innobase Oy
Created 3/26/1996 Heikki Tuuri
*******************************************************/
#include "trx0sys.h"
#ifdef UNIV_NONINL
#include "trx0sys.ic"
#endif
#include "fsp0fsp.h"
#include "mtr0mtr.h"
#include "trx0trx.h"
#include "trx0rseg.h"
#include "trx0undo.h"
#include "srv0srv.h"
#include "trx0purge.h"
#include "log0log.h"
#include "os0file.h"
/* The transaction system */
trx_sys_t* trx_sys = NULL;
trx_doublewrite_t* trx_doublewrite = NULL;
/* The following is set to TRUE when we are upgrading from the old format data
files to the new >= 4.1.x format multiple tablespaces format data files */
ibool trx_doublewrite_must_reset_space_ids = FALSE;
/* The following is TRUE when we are using the database in the new format,
i.e., we have successfully upgraded, or have created a new database
installation */
ibool trx_sys_multiple_tablespace_format = FALSE;
/* In a MySQL replication slave, in crash recovery we store the master log
file name and position here. We have successfully got the updates to InnoDB
up to this position. If .._pos is -1, it means no crash recovery was needed,
or there was no master log position info inside InnoDB. */
char trx_sys_mysql_master_log_name[TRX_SYS_MYSQL_LOG_NAME_LEN];
ib_longlong trx_sys_mysql_master_log_pos = -1;
/* If this MySQL server uses binary logging, after InnoDB has been inited
and if it has done a crash recovery, we store the binlog file name and position
here. If .._pos is -1, it means there was no binlog position info inside
InnoDB. */
char trx_sys_mysql_bin_log_name[TRX_SYS_MYSQL_LOG_NAME_LEN];
ib_longlong trx_sys_mysql_bin_log_pos = -1;
/********************************************************************
Determines if a page number is located inside the doublewrite buffer. */
ibool
trx_doublewrite_page_inside(
/*========================*/
/* out: TRUE if the location is inside
the two blocks of the doublewrite buffer */
ulint page_no) /* in: page number */
{
if (trx_doublewrite == NULL) {
return(FALSE);
}
if (page_no >= trx_doublewrite->block1
&& page_no < trx_doublewrite->block1
+ TRX_SYS_DOUBLEWRITE_BLOCK_SIZE) {
return(TRUE);
}
if (page_no >= trx_doublewrite->block2
&& page_no < trx_doublewrite->block2
+ TRX_SYS_DOUBLEWRITE_BLOCK_SIZE) {
return(TRUE);
}
return(FALSE);
}
/********************************************************************
Creates or initialializes the doublewrite buffer at a database start. */
static
void
trx_doublewrite_init(
/*=================*/
byte* doublewrite) /* in: pointer to the doublewrite buf
header on trx sys page */
{
trx_doublewrite = mem_alloc(sizeof(trx_doublewrite_t));
/* Since we now start to use the doublewrite buffer, no need to call
fsync() after every write to a data file */
#ifdef UNIV_DO_FLUSH
os_do_not_call_flush_at_each_write = TRUE;
#endif /* UNIV_DO_FLUSH */
mutex_create(&trx_doublewrite->mutex, SYNC_DOUBLEWRITE);
trx_doublewrite->first_free = 0;
trx_doublewrite->block1 = mach_read_from_4(
doublewrite + TRX_SYS_DOUBLEWRITE_BLOCK1);
trx_doublewrite->block2 = mach_read_from_4(
doublewrite + TRX_SYS_DOUBLEWRITE_BLOCK2);
trx_doublewrite->write_buf_unaligned = ut_malloc(
(1 + 2 * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE) * UNIV_PAGE_SIZE);
trx_doublewrite->write_buf = ut_align(
trx_doublewrite->write_buf_unaligned, UNIV_PAGE_SIZE);
trx_doublewrite->buf_block_arr = mem_alloc(
2 * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE * sizeof(void*));
}
/********************************************************************
Marks the trx sys header when we have successfully upgraded to the >= 4.1.x
multiple tablespace format. */
void
trx_sys_mark_upgraded_to_multiple_tablespaces(void)
/*===============================================*/
{
page_t* page;
byte* doublewrite;
mtr_t mtr;
/* We upgraded to 4.1.x and reset the space id fields in the
doublewrite buffer. Let us mark to the trx_sys header that the upgrade
has been done. */
mtr_start(&mtr);
page = buf_page_get(TRX_SYS_SPACE, TRX_SYS_PAGE_NO, RW_X_LATCH, &mtr);
#ifdef UNIV_SYNC_DEBUG
buf_page_dbg_add_level(page, SYNC_NO_ORDER_CHECK);
#endif /* UNIV_SYNC_DEBUG */
doublewrite = page + TRX_SYS_DOUBLEWRITE;
mlog_write_ulint(doublewrite + TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED,
TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED_N,
MLOG_4BYTES, &mtr);
mtr_commit(&mtr);
/* Flush the modified pages to disk and make a checkpoint */
log_make_checkpoint_at(ut_dulint_max, TRUE);
trx_sys_multiple_tablespace_format = TRUE;
}
/********************************************************************
Creates the doublewrite buffer to a new InnoDB installation. The header of the
doublewrite buffer is placed on the trx system header page. */
void
trx_sys_create_doublewrite_buf(void)
/*================================*/
{
page_t* page;
page_t* page2;
page_t* new_page;
byte* doublewrite;
byte* fseg_header;
ulint page_no;
ulint prev_page_no;
ulint i;
mtr_t mtr;
if (trx_doublewrite) {
/* Already inited */
return;
}
start_again:
mtr_start(&mtr);
page = buf_page_get(TRX_SYS_SPACE, TRX_SYS_PAGE_NO, RW_X_LATCH, &mtr);
#ifdef UNIV_SYNC_DEBUG
buf_page_dbg_add_level(page, SYNC_NO_ORDER_CHECK);
#endif /* UNIV_SYNC_DEBUG */
doublewrite = page + TRX_SYS_DOUBLEWRITE;
if (mach_read_from_4(doublewrite + TRX_SYS_DOUBLEWRITE_MAGIC)
== TRX_SYS_DOUBLEWRITE_MAGIC_N) {
/* The doublewrite buffer has already been created:
just read in some numbers */
trx_doublewrite_init(doublewrite);
mtr_commit(&mtr);
} else {
fprintf(stderr,
"InnoDB: Doublewrite buffer not found:"
" creating new\n");
if (buf_pool_get_curr_size()
< ((2 * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE
+ FSP_EXTENT_SIZE / 2 + 100)
* UNIV_PAGE_SIZE)) {
fprintf(stderr,
"InnoDB: Cannot create doublewrite buffer:"
" you must\n"
"InnoDB: increase your buffer pool size.\n"
"InnoDB: Cannot continue operation.\n");
exit(1);
}
page2 = fseg_create(TRX_SYS_SPACE, TRX_SYS_PAGE_NO,
TRX_SYS_DOUBLEWRITE
+ TRX_SYS_DOUBLEWRITE_FSEG, &mtr);
/* fseg_create acquires a second latch on the page,
therefore we must declare it: */
#ifdef UNIV_SYNC_DEBUG
buf_page_dbg_add_level(page2, SYNC_NO_ORDER_CHECK);
#endif /* UNIV_SYNC_DEBUG */
if (page2 == NULL) {
fprintf(stderr,
"InnoDB: Cannot create doublewrite buffer:"
" you must\n"
"InnoDB: increase your tablespace size.\n"
"InnoDB: Cannot continue operation.\n");
/* We exit without committing the mtr to prevent
its modifications to the database getting to disk */
exit(1);
}
fseg_header = page + TRX_SYS_DOUBLEWRITE
+ TRX_SYS_DOUBLEWRITE_FSEG;
prev_page_no = 0;
for (i = 0; i < 2 * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE
+ FSP_EXTENT_SIZE / 2; i++) {
page_no = fseg_alloc_free_page(fseg_header,
prev_page_no + 1,
FSP_UP, &mtr);
if (page_no == FIL_NULL) {
fprintf(stderr,
"InnoDB: Cannot create doublewrite"
" buffer: you must\n"
"InnoDB: increase your"
" tablespace size.\n"
"InnoDB: Cannot continue operation.\n"
);
exit(1);
}
/* We read the allocated pages to the buffer pool;
when they are written to disk in a flush, the space
id and page number fields are also written to the
pages. When we at database startup read pages
from the doublewrite buffer, we know that if the
space id and page number in them are the same as
the page position in the tablespace, then the page
has not been written to in doublewrite. */
new_page = buf_page_get(TRX_SYS_SPACE, page_no,
RW_X_LATCH, &mtr);
#ifdef UNIV_SYNC_DEBUG
buf_page_dbg_add_level(new_page, SYNC_NO_ORDER_CHECK);
#endif /* UNIV_SYNC_DEBUG */
/* Make a dummy change to the page to ensure it will
be written to disk in a flush */
mlog_write_ulint(new_page + FIL_PAGE_DATA,
TRX_SYS_DOUBLEWRITE_MAGIC_N,
MLOG_4BYTES, &mtr);
if (i == FSP_EXTENT_SIZE / 2) {
mlog_write_ulint(doublewrite
+ TRX_SYS_DOUBLEWRITE_BLOCK1,
page_no, MLOG_4BYTES, &mtr);
mlog_write_ulint(doublewrite
+ TRX_SYS_DOUBLEWRITE_REPEAT
+ TRX_SYS_DOUBLEWRITE_BLOCK1,
page_no, MLOG_4BYTES, &mtr);
} else if (i == FSP_EXTENT_SIZE / 2
+ TRX_SYS_DOUBLEWRITE_BLOCK_SIZE) {
mlog_write_ulint(doublewrite
+ TRX_SYS_DOUBLEWRITE_BLOCK2,
page_no, MLOG_4BYTES, &mtr);
mlog_write_ulint(doublewrite
+ TRX_SYS_DOUBLEWRITE_REPEAT
+ TRX_SYS_DOUBLEWRITE_BLOCK2,
page_no, MLOG_4BYTES, &mtr);
} else if (i > FSP_EXTENT_SIZE / 2) {
ut_a(page_no == prev_page_no + 1);
}
prev_page_no = page_no;
}
mlog_write_ulint(doublewrite + TRX_SYS_DOUBLEWRITE_MAGIC,
TRX_SYS_DOUBLEWRITE_MAGIC_N,
MLOG_4BYTES, &mtr);
mlog_write_ulint(doublewrite + TRX_SYS_DOUBLEWRITE_MAGIC
+ TRX_SYS_DOUBLEWRITE_REPEAT,
TRX_SYS_DOUBLEWRITE_MAGIC_N,
MLOG_4BYTES, &mtr);
mlog_write_ulint(doublewrite
+ TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED,
TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED_N,
MLOG_4BYTES, &mtr);
mtr_commit(&mtr);
/* Flush the modified pages to disk and make a checkpoint */
log_make_checkpoint_at(ut_dulint_max, TRUE);
fprintf(stderr, "InnoDB: Doublewrite buffer created\n");
trx_sys_multiple_tablespace_format = TRUE;
goto start_again;
}
}
/********************************************************************
At a database startup initializes the doublewrite buffer memory structure if
we already have a doublewrite buffer created in the data files. If we are
upgrading to an InnoDB version which supports multiple tablespaces, then this
function performs the necessary update operations. If we are in a crash
recovery, this function uses a possible doublewrite buffer to restore
half-written pages in the data files. */
void
trx_sys_doublewrite_init_or_restore_pages(
/*======================================*/
ibool restore_corrupt_pages)
{
byte* buf;
byte* read_buf;
byte* unaligned_read_buf;
ulint block1;
ulint block2;
ulint source_page_no;
byte* page;
byte* doublewrite;
ulint space_id;
ulint page_no;
ulint i;
/* We do the file i/o past the buffer pool */
unaligned_read_buf = ut_malloc(2 * UNIV_PAGE_SIZE);
read_buf = ut_align(unaligned_read_buf, UNIV_PAGE_SIZE);
/* Read the trx sys header to check if we are using the doublewrite
buffer */
fil_io(OS_FILE_READ, TRUE, TRX_SYS_SPACE, TRX_SYS_PAGE_NO, 0,
UNIV_PAGE_SIZE, read_buf, NULL);
doublewrite = read_buf + TRX_SYS_DOUBLEWRITE;
if (mach_read_from_4(doublewrite + TRX_SYS_DOUBLEWRITE_MAGIC)
== TRX_SYS_DOUBLEWRITE_MAGIC_N) {
/* The doublewrite buffer has been created */
trx_doublewrite_init(doublewrite);
block1 = trx_doublewrite->block1;
block2 = trx_doublewrite->block2;
buf = trx_doublewrite->write_buf;
} else {
goto leave_func;
}
if (mach_read_from_4(doublewrite + TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED)
!= TRX_SYS_DOUBLEWRITE_SPACE_ID_STORED_N) {
/* We are upgrading from a version < 4.1.x to a version where
multiple tablespaces are supported. We must reset the space id
field in the pages in the doublewrite buffer because starting
from this version the space id is stored to
FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID. */
trx_doublewrite_must_reset_space_ids = TRUE;
fprintf(stderr,
"InnoDB: Resetting space id's in the"
" doublewrite buffer\n");
} else {
trx_sys_multiple_tablespace_format = TRUE;
}
/* Read the pages from the doublewrite buffer to memory */
fil_io(OS_FILE_READ, TRUE, TRX_SYS_SPACE, block1, 0,
TRX_SYS_DOUBLEWRITE_BLOCK_SIZE * UNIV_PAGE_SIZE,
buf, NULL);
fil_io(OS_FILE_READ, TRUE, TRX_SYS_SPACE, block2, 0,
TRX_SYS_DOUBLEWRITE_BLOCK_SIZE * UNIV_PAGE_SIZE,
buf + TRX_SYS_DOUBLEWRITE_BLOCK_SIZE * UNIV_PAGE_SIZE,
NULL);
/* Check if any of these pages is half-written in data files, in the
intended position */
page = buf;
for (i = 0; i < TRX_SYS_DOUBLEWRITE_BLOCK_SIZE * 2; i++) {
page_no = mach_read_from_4(page + FIL_PAGE_OFFSET);
if (trx_doublewrite_must_reset_space_ids) {
space_id = 0;
mach_write_to_4(page
+ FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID, 0);
/* We do not need to calculate new checksums for the
pages because the field .._SPACE_ID does not affect
them. Write the page back to where we read it from. */
if (i < TRX_SYS_DOUBLEWRITE_BLOCK_SIZE) {
source_page_no = block1 + i;
} else {
source_page_no = block2
+ i - TRX_SYS_DOUBLEWRITE_BLOCK_SIZE;
}
fil_io(OS_FILE_WRITE, TRUE, 0, source_page_no, 0,
UNIV_PAGE_SIZE, page, NULL);
/* printf("Resetting space id in page %lu\n",
source_page_no); */
} else {
space_id = mach_read_from_4(
page + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID);
}
if (!restore_corrupt_pages) {
/* The database was shut down gracefully: no need to
restore pages */
} else if (!fil_tablespace_exists_in_mem(space_id)) {
/* Maybe we have dropped the single-table tablespace
and this page once belonged to it: do nothing */
} else if (!fil_check_adress_in_tablespace(space_id,
page_no)) {
fprintf(stderr,
"InnoDB: Warning: a page in the"
" doublewrite buffer is not within space\n"
"InnoDB: bounds; space id %lu"
" page number %lu, page %lu in"
" doublewrite buf.\n",
(ulong) space_id, (ulong) page_no, (ulong) i);
} else if (space_id == TRX_SYS_SPACE
&& ((page_no >= block1
&& page_no
< block1 + TRX_SYS_DOUBLEWRITE_BLOCK_SIZE)
|| (page_no >= block2
&& page_no
< (block2
+ TRX_SYS_DOUBLEWRITE_BLOCK_SIZE)))) {
/* It is an unwritten doublewrite buffer page:
do nothing */
} else {
/* Read in the actual page from the data files */
fil_io(OS_FILE_READ, TRUE, space_id, page_no, 0,
UNIV_PAGE_SIZE, read_buf, NULL);
/* Check if the page is corrupt */
if (buf_page_is_corrupted(read_buf)) {
fprintf(stderr,
"InnoDB: Warning: database page"
" corruption or a failed\n"
"InnoDB: file read of page %lu.\n",
(ulong) page_no);
fprintf(stderr,
"InnoDB: Trying to recover it from"
" the doublewrite buffer.\n");
if (buf_page_is_corrupted(page)) {
fprintf(stderr,
"InnoDB: Dump of the page:\n");
buf_page_print(read_buf);
fprintf(stderr,
"InnoDB: Dump of"
" corresponding page"
" in doublewrite buffer:\n");
buf_page_print(page);
fprintf(stderr,
"InnoDB: Also the page in the"
" doublewrite buffer"
" is corrupt.\n"
"InnoDB: Cannot continue"
" operation.\n"
"InnoDB: You can try to"
" recover the database"
" with the my.cnf\n"
"InnoDB: option:\n"
"InnoDB: set-variable="
"innodb_force_recovery=6\n");
exit(1);
}
/* Write the good page from the
doublewrite buffer to the intended
position */
fil_io(OS_FILE_WRITE, TRUE, space_id,
page_no, 0,
UNIV_PAGE_SIZE, page, NULL);
fprintf(stderr,
"InnoDB: Recovered the page from"
" the doublewrite buffer.\n");
}
}
page += UNIV_PAGE_SIZE;
}
fil_flush_file_spaces(FIL_TABLESPACE);
leave_func:
ut_free(unaligned_read_buf);
}
/********************************************************************
Checks that trx is in the trx list. */
ibool
trx_in_trx_list(
/*============*/
/* out: TRUE if is in */
trx_t* in_trx) /* in: trx */
{
trx_t* trx;
ut_ad(mutex_own(&(kernel_mutex)));
trx = UT_LIST_GET_FIRST(trx_sys->trx_list);
while (trx != NULL) {
if (trx == in_trx) {
return(TRUE);
}
trx = UT_LIST_GET_NEXT(trx_list, trx);
}
return(FALSE);
}
/*********************************************************************
Writes the value of max_trx_id to the file based trx system header. */
void
trx_sys_flush_max_trx_id(void)
/*==========================*/
{
trx_sysf_t* sys_header;
mtr_t mtr;
ut_ad(mutex_own(&kernel_mutex));
mtr_start(&mtr);
sys_header = trx_sysf_get(&mtr);
mlog_write_dulint(sys_header + TRX_SYS_TRX_ID_STORE,
trx_sys->max_trx_id, &mtr);
mtr_commit(&mtr);
}
/*********************************************************************
Updates the offset information about the end of the MySQL binlog entry
which corresponds to the transaction just being committed. In a MySQL
replication slave updates the latest master binlog position up to which
replication has proceeded. */
void
trx_sys_update_mysql_binlog_offset(
/*===============================*/
const char* file_name,/* in: MySQL log file name */
ib_longlong offset, /* in: position in that log file */
ulint field, /* in: offset of the MySQL log info field in
the trx sys header */
mtr_t* mtr) /* in: mtr */
{
trx_sysf_t* sys_header;
if (ut_strlen(file_name) >= TRX_SYS_MYSQL_LOG_NAME_LEN) {
/* We cannot fit the name to the 512 bytes we have reserved */
return;
}
sys_header = trx_sysf_get(mtr);
if (mach_read_from_4(sys_header + field
+ TRX_SYS_MYSQL_LOG_MAGIC_N_FLD)
!= TRX_SYS_MYSQL_LOG_MAGIC_N) {
mlog_write_ulint(sys_header + field
+ TRX_SYS_MYSQL_LOG_MAGIC_N_FLD,
TRX_SYS_MYSQL_LOG_MAGIC_N,
MLOG_4BYTES, mtr);
}
if (0 != strcmp((char*) (sys_header + field + TRX_SYS_MYSQL_LOG_NAME),
file_name)) {
mlog_write_string(sys_header + field
+ TRX_SYS_MYSQL_LOG_NAME,
(byte*) file_name, 1 + ut_strlen(file_name),
mtr);
}
if (mach_read_from_4(sys_header + field
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH) > 0
|| (offset >> 32) > 0) {
mlog_write_ulint(sys_header + field
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH,
(ulint)(offset >> 32),
MLOG_4BYTES, mtr);
}
mlog_write_ulint(sys_header + field
+ TRX_SYS_MYSQL_LOG_OFFSET_LOW,
(ulint)(offset & 0xFFFFFFFFUL),
MLOG_4BYTES, mtr);
}
/*********************************************************************
Prints to stderr the MySQL binlog info in the system header if the
magic number shows it valid. */
void
trx_sys_print_mysql_binlog_offset_from_page(
/*========================================*/
byte* page) /* in: buffer containing the trx system header page,
i.e., page number TRX_SYS_PAGE_NO in the tablespace */
{
trx_sysf_t* sys_header;
sys_header = page + TRX_SYS;
if (mach_read_from_4(sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_MAGIC_N_FLD)
== TRX_SYS_MYSQL_LOG_MAGIC_N) {
fprintf(stderr,
"ibbackup: Last MySQL binlog file position %lu %lu,"
" file name %s\n",
(ulong) mach_read_from_4(
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH),
(ulong) mach_read_from_4(
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_LOW),
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_NAME);
}
}
/*********************************************************************
Stores the MySQL binlog offset info in the trx system header if
the magic number shows it valid, and print the info to stderr */
void
trx_sys_print_mysql_binlog_offset(void)
/*===================================*/
{
trx_sysf_t* sys_header;
mtr_t mtr;
ulint trx_sys_mysql_bin_log_pos_high;
ulint trx_sys_mysql_bin_log_pos_low;
mtr_start(&mtr);
sys_header = trx_sysf_get(&mtr);
if (mach_read_from_4(sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_MAGIC_N_FLD)
!= TRX_SYS_MYSQL_LOG_MAGIC_N) {
mtr_commit(&mtr);
return;
}
trx_sys_mysql_bin_log_pos_high = mach_read_from_4(
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH);
trx_sys_mysql_bin_log_pos_low = mach_read_from_4(
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_LOW);
trx_sys_mysql_bin_log_pos
= (((ib_longlong)trx_sys_mysql_bin_log_pos_high) << 32)
+ (ib_longlong)trx_sys_mysql_bin_log_pos_low;
ut_memcpy(trx_sys_mysql_bin_log_name,
sys_header + TRX_SYS_MYSQL_LOG_INFO
+ TRX_SYS_MYSQL_LOG_NAME, TRX_SYS_MYSQL_LOG_NAME_LEN);
fprintf(stderr,
"InnoDB: Last MySQL binlog file position %lu %lu,"
" file name %s\n",
trx_sys_mysql_bin_log_pos_high, trx_sys_mysql_bin_log_pos_low,
trx_sys_mysql_bin_log_name);
mtr_commit(&mtr);
}
/*********************************************************************
Prints to stderr the MySQL master log offset info in the trx system header if
the magic number shows it valid. */
void
trx_sys_print_mysql_master_log_pos(void)
/*====================================*/
{
trx_sysf_t* sys_header;
mtr_t mtr;
mtr_start(&mtr);
sys_header = trx_sysf_get(&mtr);
if (mach_read_from_4(sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_MAGIC_N_FLD)
!= TRX_SYS_MYSQL_LOG_MAGIC_N) {
mtr_commit(&mtr);
return;
}
fprintf(stderr,
"InnoDB: In a MySQL replication slave the last"
" master binlog file\n"
"InnoDB: position %lu %lu, file name %s\n",
(ulong) mach_read_from_4(sys_header
+ TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH),
(ulong) mach_read_from_4(sys_header
+ TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_LOW),
sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_NAME);
/* Copy the master log position info to global variables we can
use in ha_innobase.cc to initialize glob_mi to right values */
ut_memcpy(trx_sys_mysql_master_log_name,
sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_NAME,
TRX_SYS_MYSQL_LOG_NAME_LEN);
trx_sys_mysql_master_log_pos
= (((ib_longlong) mach_read_from_4(
sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_HIGH)) << 32)
+ ((ib_longlong) mach_read_from_4(
sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
+ TRX_SYS_MYSQL_LOG_OFFSET_LOW));
mtr_commit(&mtr);
}
/********************************************************************
Looks for a free slot for a rollback segment in the trx system file copy. */
ulint
trx_sysf_rseg_find_free(
/*====================*/
/* out: slot index or ULINT_UNDEFINED if not found */
mtr_t* mtr) /* in: mtr */
{
trx_sysf_t* sys_header;
ulint page_no;
ulint i;
ut_ad(mutex_own(&(kernel_mutex)));
sys_header = trx_sysf_get(mtr);
for (i = 0; i < TRX_SYS_N_RSEGS; i++) {
page_no = trx_sysf_rseg_get_page_no(sys_header, i, mtr);
if (page_no == FIL_NULL) {
return(i);
}
}
return(ULINT_UNDEFINED);
}
/*********************************************************************
Creates the file page for the transaction system. This function is called only
at the database creation, before trx_sys_init. */
static
void
trx_sysf_create(
/*============*/
mtr_t* mtr) /* in: mtr */
{
trx_sysf_t* sys_header;
ulint slot_no;
page_t* page;
ulint page_no;
ulint i;
ut_ad(mtr);
/* Note that below we first reserve the file space x-latch, and
then enter the kernel: we must do it in this order to conform
to the latching order rules. */
mtr_x_lock(fil_space_get_latch(TRX_SYS_SPACE), mtr);
mutex_enter(&kernel_mutex);
/* Create the trx sys file block in a new allocated file segment */
page = fseg_create(TRX_SYS_SPACE, 0, TRX_SYS + TRX_SYS_FSEG_HEADER,
mtr);
ut_a(buf_frame_get_page_no(page) == TRX_SYS_PAGE_NO);
#ifdef UNIV_SYNC_DEBUG
buf_page_dbg_add_level(page, SYNC_TRX_SYS_HEADER);
#endif /* UNIV_SYNC_DEBUG */
mlog_write_ulint(page + FIL_PAGE_TYPE, FIL_PAGE_TYPE_TRX_SYS,
MLOG_2BYTES, mtr);
/* Reset the doublewrite buffer magic number to zero so that we
know that the doublewrite buffer has not yet been created (this
suppresses a Valgrind warning) */
mlog_write_ulint(page + TRX_SYS_DOUBLEWRITE
+ TRX_SYS_DOUBLEWRITE_MAGIC, 0, MLOG_4BYTES, mtr);
sys_header = trx_sysf_get(mtr);
/* Start counting transaction ids from number 1 up */
mlog_write_dulint(sys_header + TRX_SYS_TRX_ID_STORE,
ut_dulint_create(0, 1), mtr);
/* Reset the rollback segment slots */
for (i = 0; i < TRX_SYS_N_RSEGS; i++) {
trx_sysf_rseg_set_space(sys_header, i, ULINT_UNDEFINED, mtr);
trx_sysf_rseg_set_page_no(sys_header, i, FIL_NULL, mtr);
}
/* The remaining area (up to the page trailer) is uninitialized. */
/* Create the first rollback segment in the SYSTEM tablespace */
page_no = trx_rseg_header_create(TRX_SYS_SPACE, ULINT_MAX, &slot_no,
mtr);
ut_a(slot_no == TRX_SYS_SYSTEM_RSEG_ID);
ut_a(page_no != FIL_NULL);
mutex_exit(&kernel_mutex);
}
/*********************************************************************
Creates and initializes the central memory structures for the transaction
system. This is called when the database is started. */
void
trx_sys_init_at_db_start(void)
/*==========================*/
{
trx_sysf_t* sys_header;
ib_longlong rows_to_undo = 0;
const char* unit = "";
trx_t* trx;
mtr_t mtr;
mtr_start(&mtr);
ut_ad(trx_sys == NULL);
mutex_enter(&kernel_mutex);
trx_sys = mem_alloc(sizeof(trx_sys_t));
sys_header = trx_sysf_get(&mtr);
trx_rseg_list_and_array_init(sys_header, &mtr);
trx_sys->latest_rseg = UT_LIST_GET_FIRST(trx_sys->rseg_list);
/* VERY important: after the database is started, max_trx_id value is
divisible by TRX_SYS_TRX_ID_WRITE_MARGIN, and the 'if' in
trx_sys_get_new_trx_id will evaluate to TRUE when the function
is first time called, and the value for trx id will be written
to the disk-based header! Thus trx id values will not overlap when
the database is repeatedly started! */
trx_sys->max_trx_id = ut_dulint_add(
ut_dulint_align_up(mtr_read_dulint(
sys_header
+ TRX_SYS_TRX_ID_STORE, &mtr),
TRX_SYS_TRX_ID_WRITE_MARGIN),
2 * TRX_SYS_TRX_ID_WRITE_MARGIN);
UT_LIST_INIT(trx_sys->mysql_trx_list);
trx_lists_init_at_db_start();
if (UT_LIST_GET_LEN(trx_sys->trx_list) > 0) {
trx = UT_LIST_GET_FIRST(trx_sys->trx_list);
for (;;) {
if ( trx->conc_state != TRX_PREPARED) {
rows_to_undo += ut_conv_dulint_to_longlong(
trx->undo_no);
}
trx = UT_LIST_GET_NEXT(trx_list, trx);
if (!trx) {
break;
}
}
if (rows_to_undo > 1000000000) {
unit = "M";
rows_to_undo = rows_to_undo / 1000000;
}
fprintf(stderr,
"InnoDB: %lu transaction(s) which must be"
" rolled back or cleaned up\n"
"InnoDB: in total %lu%s row operations to undo\n",
(ulong) UT_LIST_GET_LEN(trx_sys->trx_list),
(ulong) rows_to_undo, unit);
fprintf(stderr, "InnoDB: Trx id counter is %lu %lu\n",
(ulong) ut_dulint_get_high(trx_sys->max_trx_id),
(ulong) ut_dulint_get_low(trx_sys->max_trx_id));
}
UT_LIST_INIT(trx_sys->view_list);
trx_purge_sys_create();
mutex_exit(&kernel_mutex);
mtr_commit(&mtr);
}
/*********************************************************************
Creates and initializes the transaction system at the database creation. */
void
trx_sys_create(void)
/*================*/
{
mtr_t mtr;
mtr_start(&mtr);
trx_sysf_create(&mtr);
mtr_commit(&mtr);
trx_sys_init_at_db_start();
}