problems with concurrency on SMP systems because of word packing issues.
The number of trx_t's in a system is not sufficient enough to require that
we try and save a few bytes in the data structure.

See rb://255 for details.

is held by the caller.

ChangeLog entry for r6674.

This is a backport of r4924.
mem_heap_get_size() scans all allocated blocks to calculate the total
size of the heap. This patch introduces a new, total_size, field in
mem_block_info_struct. This field is valid only for base block 
(i.e.: the first block allocated for the heap) and is set to
ULINT_UNDEFINED in other blocks.
This considerably improves the performance of redo scan during recovery.

rb://108 issue#216

Approved by: Heikki

------------------------------------------------------------------------

When rolling back an incomplete transaction in purge, tolerate missing
BLOBs also in update undo, when undoing an INSERT by updating a delete-marked
record, and the delete-marked record is no longer needed.
Previously, we only tolerated missing BLOBs in insert undo.
This merely fixes a debug assertion; the code performed correctly
without UNIV_DEBUG.

rb://249 approved by Sunny Bains.

row_undo_mod_remove_clust_low(): Augment the function comment.

row_undo_mod_remove_clust_low(), row_undo_mod_del_mark_or_remove_sec_low(),
row_undo_mod_del_mark_or_remove_sec(), row_undo_mod_upd_del_sec():
Add ut_ad(node->rec_type == TRX_UNDO_UPD_DEL_REC);

btr_pcur_open_with_no_init() is a macro, do not mix preprocessor
directives in the macro invocation, because it is implementation-defined
whether that is going to work.

  ------------------------------------------------------------------------
  r6545 | jyang | 2010-02-03 03:57:32 +0200 (Wed, 03 Feb 2010) | 8 lines
  Changed paths:
     M /branches/5.1/lock/lock0lock.c

  branches/5.1: Fix bug #49001, "SHOW INNODB STATUS deadlock info
  incorrect when deadlock detection aborts". Print the correct
  lock owner when recursive function lock_deadlock_recursive()
  exceeds its maximum depth LOCK_MAX_DEPTH_IN_DEADLOCK_CHECK.

  rb://217, approved by Marko.
  ------------------------------------------------------------------------
  r6613 | inaam | 2010-02-09 20:23:09 +0200 (Tue, 09 Feb 2010) | 11 lines
  Changed paths:
     M /branches/5.1/buf/buf0buf.c
     M /branches/5.1/buf/buf0rea.c
     M /branches/5.1/include/buf0rea.h

  branches/5.1: Fix Bug #38901
  InnoDB logs error repeatedly when trying to load page into buffer pool

  In buf_page_get_gen() if we are unable to read a page (because of
  corruption or some other reason) we keep on retrying. This fills up
  error log with millions of entries in no time and we'd eventually run
  out of disk space. This patch limits the number of attempts that we
  make (currently set to 100) and after that we abort with a message.

  rb://241 Approved by: Heikki
  ------------------------------------------------------------------------

Let the master thread sleep if the amount of work to be done is
calibrated as taking less than a second.

Approved by: Heikki

Drop the temporary tables and indexes after enabling sync order checks.
This should not make any difference. This could have been done in r6611.

innobase_start_or_create_for_mysql(): Roll back data dictionary
transactions before scanning the *.ibd files. Then, data dictionary
records can be loaded to the cache before opening the *.ibd files.

recv_recovery_rollback_active(): Refactored from
recv_recovery_from_checkpoint_finish().

rb://235, committing without review, because this is needed for
TablespaceDictionary.

first load them to the data dictionary cache and use the normal
routines for dropping tables or indexes.  This should reduce the
risk of bugs and also make the code compatible with the upcoming
TablespaceDictionary implementation.

DICT_SYS_INDEXES_NAME_FIELD: The clustered index position of SYS_INDEXES.NAME.

row_merge_drop_temp_indexes(): Scan SYS_INDEXES for tables containing
temporary indexes, and load the tables as needed. Invoke
row_merge_drop_index() to drop the indexes.

row_mysql_drop_temp_tables(): Scan SYS_TABLES for temporary tables,
load them with dict_load_table() and drop them with
row_drop_table_for_mysql().

rb://251, not yet reviewed

Replace if (!cond) ut_error; tests with ut_a(cond).

on SYS_INDEXES.TABLE_ID.  INDEX_ID must be instance-widely unique,
because SYS_FIELDS is not indexed by TABLE_ID.

to check if the joining transaction has any other transactions waiting on
its locks. This optimization results in excessive deadlocks when running
Sysbench with a large number of threads. The function seems to return
FALSE positives.

rb://250

b-tree cursor functions to the buffer pool requests, in order to make
the latch diagnostics more accurate.

buf_page_optimistic_get_func(): Renamed to buf_page_optimistic_get().

btr_page_get_father_node_ptr(), btr_insert_on_non_leaf_level(),
btr_pcur_open(), btr_pcur_open_with_no_init(), btr_pcur_open_on_user_rec(),
btr_pcur_open_at_rnd_pos(), btr_pcur_restore_position(),
btr_cur_open_at_index_side(), btr_cur_open_at_rnd_pos():
Rename the function to _func and add the parameters file, line.
Define wrapper macros with __FILE__, __LINE__.

btr_cur_search_to_nth_level(): Add the parameters file, line.

is not relocated when freeing a compressed block.  This avoids the
costly rescan of the LRU list.  (Bug #35077, Issue #449)

At most one buffer-fix will be active at a time, affecting two blocks:
the buf_page_t and the compressed page frame. This should not block
the memory defragmentation in buf0buddy.c too much.  In fact, it may
avoid unnecessary copying if also prev_bpage belongs to the tablespace
that is being invalidated.

rb://240

ha_innobase::change_active_index(): Clean up code formatting.

ha_innobase::check(): Incorporate the code from
row_check_table_for_mysql().  Report errors to the client connection
instead of writing them to the error log.

row_check_table_for_mysql(): Remove.

row_check_index_for_mysql(): Renamed from row_scan_and_check_index().
Let the caller initialize prebuilt, and assume that the index is usable.

rb://178 approved by Sunny Bains

in InnoDB must be comparable with that of MySQL to tolerate
possible dictionary inconsistency. Fix Mantis issue #455,
"UNIV_DEBUG+ assert ha_innodb.cc:3152 ib_num_index >= 
mysql_num_index".

rb://248 Approved by Marko.
 

  ------------------------------------------------------------------------
  r6488 | sunny | 2010-01-21 02:55:08 +0200 (Thu, 21 Jan 2010) | 2 lines
  Changed paths:
     M /branches/5.1/mysql-test/innodb-autoinc.result
     M /branches/5.1/mysql-test/innodb-autoinc.test

  branches/5.1: Factor out test for bug#44030 from innodb-autoinc.test
  into a separate test/result files.
  ------------------------------------------------------------------------
  r6489 | sunny | 2010-01-21 02:57:50 +0200 (Thu, 21 Jan 2010) | 2 lines
  Changed paths:
     A /branches/5.1/mysql-test/innodb-autoinc-44030.result
     A /branches/5.1/mysql-test/innodb-autoinc-44030.test

  branches/5.1: Factor out test for bug#44030 from innodb-autoinc.test
  into a separate test/result files.
  ------------------------------------------------------------------------
  r6492 | sunny | 2010-01-21 09:38:35 +0200 (Thu, 21 Jan 2010) | 1 line
  Changed paths:
     M /branches/5.1/mysql-test/innodb-autoinc-44030.test

  branches/5.1: Add reference to bug#47621 in the comment.
  ------------------------------------------------------------------------
  r6535 | sunny | 2010-01-30 00:08:40 +0200 (Sat, 30 Jan 2010) | 11 lines
  Changed paths:
     M /branches/5.1/handler/ha_innodb.cc

  branches/5.1: Undo the change from r6424. We need to return DB_SUCCESS even
  if we were unable to initialize the tabe autoinc value. This is required for
  the open to succeed. The only condition we currently treat as a hard error
  is if the autoinc field instance passed in by MySQL is NULL.

  Previously if the table autoinc value was 0 and the next value was requested
  we had an assertion that would fail. Change that assertion and treat a value
  of 0 to mean that the autoinc system is unavailable. Generation of next
  value will now return failure.

  rb://237
  ------------------------------------------------------------------------
  r6536 | sunny | 2010-01-30 00:13:42 +0200 (Sat, 30 Jan 2010) | 6 lines
  Changed paths:
     M /branches/5.1/handler/ha_innodb.cc
     M /branches/5.1/mysql-test/innodb-autoinc.result
     M /branches/5.1/mysql-test/innodb-autoinc.test

  branches/5.1: Check *first_value everytime against the column max
  value and  set *first_value to next autoinc if it's > col max value.
  ie.  not rely on what is passed in from MySQL.

  [49497] Error 1467 (ER_AUTOINC_READ_FAILED) on inserting a negative value
  rb://236
  ------------------------------------------------------------------------
  r6537 | sunny | 2010-01-30 00:35:00 +0200 (Sat, 30 Jan 2010) | 2 lines
  Changed paths:
     M /branches/5.1/handler/ha_innodb.cc
     M /branches/5.1/mysql-test/innodb-autoinc.result
     M /branches/5.1/mysql-test/innodb-autoinc.test

  branches/5.1: Undo r6536.
  ------------------------------------------------------------------------
  r6538 | sunny | 2010-01-30 00:43:06 +0200 (Sat, 30 Jan 2010) | 6 lines
  Changed paths:
     M /branches/5.1/handler/ha_innodb.cc
     M /branches/5.1/mysql-test/innodb-autoinc.result
     M /branches/5.1/mysql-test/innodb-autoinc.test

  branches/5.1: Check *first_value every time against the column max
  value and  set *first_value to next autoinc if it's > col max value.
  ie.  not rely on what is passed in from MySQL.

  [49497] Error 1467 (ER_AUTOINC_READ_FAILED) on inserting a negative value
  rb://236
  ------------------------------------------------------------------------

1. First scan the joining transaction's locks and check if no other
transaction is waiting for a lock held by the joining transaction.
If no other transaction is waiting then  no deadlock an occur and
we avoid doing an exhaustive search.

2. Change the direction of the lock traversal from backward to forward.
Previously we traversed backward from the lock that has to wait, the function
to that fetched the previous node was very inefficient resulting in O(n^2)
access to the rec lock list.

Fix Bug #49047 InnoDB deadlock detection is CPU intensive with many locks on a single row.

rb://218

Extend the comment about row_mysql_handle_errors().

Suggested by:	Heikki

number to InnoDB index structure directly. Fix Bug #47622:
"the new index is added before the existing ones in MySQL,
but after one in SE".

rb://215, approved by Marko

acquire the block_mutex for every block in the LRU list. Only acquire
it when holding buf_pool_mutex is not sufficient. This should speed up
the function and considerably reduce traffic on the memory bus and
caches.

I noticed this deficiency when working on Issue #157.
This deficiency popped up again in Issue #449 (Bug #35077),
which this fix does not fully address.

rb://78 revision 1 approved by Heikki Tuuri.

This addresses the third aspect of Bug #41609.

row_mysql_drop_temp_tables(): New function, to drop all temporary
tables. These can be distinguished by the least significant bit of
MIX_LEN. However, we will skip ROW_FORMAT=REDUNDANT tables, because in
the records for those tables, that bit may be garbage.

recv_recovery_from_checkpoint_finish(): Invoke
row_mysql_drop_temp_tables().  Normally, if the .frm files for the
temporary tables exist at startup, MySQL will ask InnoDB to drop the
temporary tables.  However, if the files are deleted, for instance, by
the boot scripts of the operating system, the tables would remain in
the InnoDB data dictionary unless someone digs them up by
innodb_table_monitor and creates .frm files for dropping the tables.

rb://221 approved by Sunny Bains.

and do not call ibuf_merge_or_delete_for_page() in crash recovery,
before the redo log has been applied.
This could cure some hard-to-repeat, hard-to-explain bugs
related to secondary indexes.

A possible recipe to repeat the bug:

1. update a secondary index leaf page on a compressed table
2. evict the page from the buffer pool while it is still dirty
3. ibuf_insert() something for the page
4. crash
5. crash recovery; ibuf merge would be done too early,
before applying redo log to the sec index page or the ibuf pages

embedded mode

This is 2nd part of the fix for bug#49396. The 1st part is
innodb.test. Tested in both embedded mode and normal server mode.

to pick up the first part fix of bug49396.

    ------------------------------------------------------------------------
    r6471 | calvin | 2010-01-15 17:43:27 -0600 (Fri, 15 Jan 2010) | 4 lines

    branches/5.1: fix bug#49396: main.innodb test fails in embedded mode

    Change replace_result by using $MYSQLD_DATADIR. Tested in both embedded
    mode and normal server mode.
    ------------------------------------------------------------------------

log_sys->written_to_all_lsn does not accurately represent the LSN
upto which write and flush has taken place. Under a race condition
it can fall behind log_sys->flushed_to_disk_lsn which is accurate.
Besides written_to_all_lsn is redundant as currently InnoDB supports
only one log group.

Approved by: Heikki

Update PAGE_MAX_TRX_ID before attempting to compress the page.  This
fixes Issue #382 (a debug assertion failure in page_zip_reorganize())
and reduces the generated redo log.  There was no bug or crash in
non-debug builds.

queues when the thread is not holding a space->latch.

When UNIV_DEBUG is defined while UNIV_SYNC_DEBUG is not,
latching order violations will still occur and deadlocks will be possible.

sync_thread_levels_nonempty_gen(): Renamed from
sync_thread_levels_empty_gen().  Return the violating latch or NULL
instead of FALSE or TRUE, except that there will be a ut_error before
the non-NULL return.

sync_thread_levels_empty_gen(): A macro that negates the return value of
sync_thread_levels_nonempty_gen().

sync_thread_levels_contains(): New function, based on
sync_thread_levels_nonempty_gen().

This should fix Issue #441.