The method Item_ref::not_null_tables() returned incorrect bitmap
for outer references to view columns. This could cause an invalid
conversion of an outer join into an inner join that could lead
to a wrong result set for a query with a correlated subquery over
an outer join whose where condition had an outer reference to a view.

The method Item_func_isnull::update_used_tables() erroneously did not
update cached values stored in the fields used_tables_cache and
const_item_cache of the Item_func_isnull objects. As a result the
Item_func_isnull::used_tables() returned wrong bitmaps and, as a
consequence, push-down predicates could be attached to wrong tables.

Identified all test cases in the MySQL file subquery_mat.inc that are
not present in MariaDB. In total found 8 test cases for the following
MySQL bugs:
* BUG#49630 - not a bug in MariaDB, added test case
* BUG#52538 - not a bug in MariaDB, added test case (checked with VG)
* BUG#53103 - not a bug in MariaDB, added test case
* BUG#54511 - not a bug in MariaDB, added test case
* BUG#56367 - not a bug in MariaDB, added test case
* BUG#59833 - not a bug in MariaDB, added test case
* BUG#11852644 - not a bug in MariaDB, added test case
* BUG#12668294 - not a bug in MariaDB, added test case

All of these MySQL bugs are not present in MariaDB 5.3.

The comparison was based on the following version of
mysql-trunk:

revno: 3350 [merge]
committer: Marko Mäkelä <marko.makela@oracle.com>
branch nick: mysql-trunk
timestamp: Mon 2011-08-08 12:42:09 +0300
message:
  Merge mysql-5.5 to mysql-trunk.

This bug is a special case of lp:813447.

Analysis:
Constant optimization finds that the condition t2.a = 1
can be used to access the primary key of table 't2'. As
a result both outer table t1,t2 are considered as constant
when we reach the execution phase. At the same time, during
constant optimization, the IN predicate is not evaluated
because it is expensive.

When execution of the outer query reaches do_select(),
control flow enter the branch:
if (join->table_count == join->const_tables)
{ ... }
This branch checks only the WHERE and HAVING clauses,
but doesn't check the ON clauses of the query. Since the
IN predicate was not evaluated during optimization, it is
not evaluated at all, thus execution doesn't detect that
the ON clause is FALSE.

Solution:
Similar to the patch for bug lp:813447, exclude system
tables from constant substitution based on unique key
lookups if there is an expensive ON condition on the
inner table.

Do not optimize derived table for the second time ever.

- create_ref_for_key() has the code that walks KEYUSE array and tries to use
  maximum number of keyparts for ref (and eq_ref and ref_or_null) access.
  When one constructs ref access for table that is inside a SJ-Materialization
  nest, it is not possible to use tables that are ouside the nest (because 
  materialization is performed before they have any "current value").
  The bug was caused by this function not taking this into account.

The reason for the long shutdown is hanging in io threads. It appears
that just closing completion port on XP does not necessarily signal 
thread waiting in GetIOCompletionStatus() (even if this works fine
on later Windows versions)

The fix is to wakeup background threads using PostQueuedCompletionStatus()
with a special 'key' parameter indicating shutdown.

The reason for the crash is Innodb assertion after trying to load condition variables function
dynamically and not finding them

The fix is to skip dynamic loading if srv_use_native_conditions is FALSE. srv_use_native_conditions 
is derived from Windows version and would be FALSE on XP and TRUE on later Windows.

This is the same handling as in MySQL 5.. In Maria 5.3 srv_use_native_conditions check was 
presumably  lost in the downporting.

revno: 2876.47.174
revision-id: jorgen.loland@oracle.com-20110519120355-qn7eprkad9jqwu5j
parent: mayank.prasad@oracle.com-20110518143645-bdxv4udzrmqsjmhq
committer: Jorgen Loland <jorgen.loland@oracle.com>
branch nick: mysql-trunk-11765831
timestamp: Thu 2011-05-19 14:03:55 +0200
message:
  BUG#11765831: 'RANGE ACCESS' MAY INCORRECTLY FILTER 
                        AWAY QUALIFYING ROWS
        
  The problem was that the ranges created when OR'ing two 
  conditions could be incorrect. Without the bugfix, 
  "I <> 6 OR (I <> 8 AND J = 5)" would create these ranges:
  
  "NULL < I < 6",
  "6 <= I <= 6 AND 5 <= J <= 5",
  "6 < I < 8",
  "8 <= I <= 8 AND 5 <= J <= 5",
  "8 < I"
  
  While the correct ranges is
  "NULL < I < 6",
  "6 <= I <= 6 AND 5 <= J <= 5",
  "6 < I"
  
  The problem occurs when key_or() ORs
  (1) "NULL < I < 6, 6 <= I <= 6 AND 5 <= J <= 5, 6 < I" with 
  (2) "8 < I AND 5 <= J <= 5"
  
  The reason for the bug is that in key_or(), SEL_ARG *tmp is 
  used to point to the range in (1) above that is merged with 
  (2) while key1 points to the root of the red-black tree of 
  (1). When merging (1) and (2), tmp refers to the "6 < I" 
  part whereas the root is the "6 <= ... AND 5 <= J <= 5" part. 
  
  key_or() decides that the tmp range needs to be split into
  "6 < I < 8, 8 <= I <= 8, 8 < I", in which next_key_part of the 
  second range should be that of tmp. However, next_key_part is
  set to key1->next_key_part ("5 <= J <= 5") instead of 
  tmp->next_key_part (empty). Fixing this gives the correct but
  not optimal ranges:
  "NULL < I < 6",
  "6 <= I <= 6 AND 5 <= J <= 5",
  "6 < I < 8",
  "8 <= I <= 8",
  "8 < I"
  
  A second problem can be seen above: key_or() may create 
  adjacent ranges that could be replaced with a single range. 
  Fixes for this is also included in the patch so that the range
  above becomes correct AND optimal:
  "NULL < I < 6",
  "6 <= I <= 6 AND 5 <= J <= 5",
  "6 < I"
  
  Merging adjacent ranges like this gives a slightly lower cost 
  estimate for the range access.

revno: 3363.3.16
revision-id: jorgen.loland@oracle.com-20110506132631-5wickj6dvrh1dpj6
parent: alexander.nozdrin@oracle.com-20110506132138-46459va9vcbd4nz0
committer: Jorgen Loland <jorgen.loland@oracle.com>
branch nick: mysql-trunk-11765831
timestamp: Fri 2011-05-06 15:26:31 +0200
message:
  BUG#11765831: 'RANGE ACCESS' MAY INCORRECTLY FILTER
                AWAY QUALIFYING ROWS

  Preparation patch (does not include fix for the bug):

   * Extensively document key_or()
   * Remove tab indentations from key_or()
   * Minor code changes like using existing utility functions
     in key_or()

LPBug#819711 : optimizer_switch must be reported on segfault
LPBug#820169:  Full query text must be reported on crash

Problem was the parsing of test suite files for various tags and options.
This was done inefficiently, and include files were re-parsed for every
place they were included. This caused a delay of 20 seconds or so before
the first test started to run.

By parsing more efficiently and re-using first parse for subsequent
inclusion of the same file, time spent parsing is reduced to less than
1 second, and start appears instantaneous.

(With this patch, full ./mtr runs in 3 minutes on my laptop (release
build.)

mysql-test/suite/innodb_plugin/t/innodb_bug52663.test:
  Test is fairly slow, so try to avoid getting stuck with it at the end
  while other workers are idle.

This problem could be observed for queries with nested outer joins
for which the not_exist optimization were applicable. 
The problem was caused by the code of the patch for bug #49322
that erroneously forced the return to the previous nested loop
level when the join algorithm successfully builds a partial record
for an embedded outer to which the not_exist optimization could be
applied.
Actually the immediate return to the previous nested loops level
is correct only if this partial record is rejected by a predicate
pushed down to one of the inner tables of this outer join. Otherwise
attempts to find extensions of this record must be made.

sql/sql_expression_cache.cc:
  Do not go on disk if hit rate is not good.
  Local hit/miss counters added.
sql/sql_expression_cache.h:
  Local hit/miss counters added.

Fixed lp:814237: Wrong mutex usage in Aria


storage/maria/ma_bitmap.c:
  Added call to _ma_bitmap_mark_file_changed() when flushing bitmap. This fixes lp:814237

scripts/mysqldumpslow.sh:
  Remove not used --basedir option
  Ensure that the last --datadir option is used from the my.cnf files.

The bug happens when one uses MAX_ROWS=# with Aria & row_format=page and one insert more than # rows.


mysql-test/mysql-test-run.pl:
  Ignore table is full error messages
mysql-test/suite/maria/r/max_length.result:
  Test case for 'Table is full'
mysql-test/suite/maria/t/max_length.test:
  Test case for 'Table is full'
storage/maria/ma_bitmap.c:
  Ensure that we don't allocate bits outside of max_data_file_size.
  Adjust max_data_file_size based on bitmap alignments.
  Backport fix to adjust wrong first_bitmap_with_space.
storage/maria/ma_blockrec.c:
  Calculate value of max_data_file_length
storage/maria/ma_blockrec.h:
  Updated prototype for _ma_bitmap_init()
storage/maria/ma_check.c:
  Give warnings if file sizes are above max file sizes.
  Give more warnings in case of errors.
  Have maria_chk write if table is recreated.
storage/maria/ma_create.c:
  Better calculation of max_data_file_length and thus data pointer length.
  Fixes some wrong pointer lengths when using MAX_ROWS=#
storage/maria/ma_open.c:
  Removed duplicate assigment.
  Use block size from file instead of global variable.
storage/maria/maria_chk.c:
  Remove -1 from printed file length
storage/maria/maria_def.h:
  Update struct st_maria_file_bitmap

the optimizer switch.

off by default.

In case of two views with subqueries it is dificult to decide about order of injected ORDER BY clauses.
A simple solution is just prohibit ORDER BY injection if there is other order by.

mysql-test/r/view.result:
  New test added, old test changed.
mysql-test/t/view.test:
  New test aded.
sql/share/errmsg.txt:
  new warning added.
sql/sql_view.cc:
  Inject ORDER BY only if there is no other one.
  
  Warning about ignoring ORDER BY in this case for EXPLAIN EXTENDED.

- Correct handling of outer joins + DuplicateWeedout (docs pending)

Fixed explains of previous patch.

mysql-test/r/explain.result:
  Fixed explains of previous patch.
mysql-test/r/join_outer.result:
  Fixed explains of previous patch.
mysql-test/r/negation_elimination.result:
  Fixed explains of previous patch.
mysql-test/r/view.result:
  Fixed explains of previous patch.
mysql-test/suite/innodb/r/innodb_mysql.result:
  Removed duplicate test suite.
mysql-test/suite/innodb/t/innodb_mysql.test:
  Removed duplicate test suite.
mysql-test/suite/innodb_plugin/r/innodb_mysql.result:
  Removed duplicate test suite.
mysql-test/suite/innodb_plugin/t/innodb_mysql.test:
  Removed duplicate test suite.
sql/opt_range.h:
  Removed incorrect fix.
sql/records.cc:
  Removed incorrect fix.

There are 2 volatile condition constructions AND/OR constructions and fields(references) when first
good supported to be top elements of conditions because it is normal practice
(see copy_andor_structure for example) fields without any expression in the condition is really rare
and mostly useless case however it could lead to problems when optimiser changes/moves them unaware
of other variables referring to them. An easy solution of this problem is just to replace single field
in a condition with equivalent expression well supported by the server (<field> -> <field> != 0).

mysql-test/r/view.result:
  New test added.
mysql-test/t/view.test:
  New test added.
sql/sql_parse.cc:
  <field> -> <field> != 0
sql/sql_yacc.yy:
  <field> -> <field> != 0