revno: 2630.22.41
committer: Alexander Nozdrin <alik@mysql.com>
branch nick: 6.0-rt-bug39255
timestamp: Thu 2008-10-16 16:39:30 +0400
message:
  A patch for Bug#39255: Stored procedures: crash if function
  references nonexistent table.
  
  The problem is not reproduced in 6.0. Adding a test case.

           redefining trigger
      
The 'table->auto_increment_field_not_null' flag is only valid within
processing of a single row, and should be set to FALSE before
navigating to the next row, or exiting the operation.
      
This bug was caused by an SQL error occuring while executing a trigger
after the flag had been set, so the normal resetting was bypassed.
The table object was then returned to the table share's cache in
a dirty condition.   When the table object was reused, an assert
caught that the flag was set.
      
This patch explicitly clears the flag on error/abort.


Backported from mysql-6.0-codebase  revid: 2617.52.1

The fix for Bug #38124 introuced a bug. If the value given 
for a set_var exceeded the length of the temporary buffer,
we would read behind the end of the buffer.  Using 
c_ptr_safe(), instead of c_ptr(), ensures that we won't
read beyond the buffer limit

mysql-6.0-codebase revid: 2617.44.1

                get_table_share, drop_open_table
            
In the partition handler code, LOCK_open and share->LOCK_ha_data
are acquired in the wrong order in certain cases.  When doing a
multi-row INSERT (i.e a INSERT..SELECT) in a table with auto-
increment column(s). the increments must be in a monotonically
continuous increasing sequence (i.e it can't have "holes"). To
achieve this, a lock is held for the duration of the operation.
share->LOCK_ha_data was used for this purpose.
            
Whenever there was a need to open a view _during_ the operation
(views are not currently pre-opened the way tables are), and
LOCK_open was grabbed, a deadlock could occur.  share->LOCK_ha_data
is other places used _while_ holding LOCK_open.
            
A new mutex was introduced in the HA_DATA_PARTITION structure,
for exclusive use of the autoincrement data fields, so we don't
need to overload the use of LOCK_ha_data here.
            
A module test case has not been supplied, since the problem occurs
as a result of a race condition, and testing for this condition 
is thus not deterministic.   Testing for it could be done by
setting up a test case as described in the bug report.

                  
When assigning the new string value to the variable, the
Item::str_value member was used.  This is not according to
the protocol.  str_value is an internal member used for
temporary assignments, and is not consistently set for all
string operations.  It is set for constant strings, so it would
work in these cases, but not for string functions (concat,
substr, etc.)
                  
The correct approach is to use Item::val_str(..) to evaluate
and retrieve the string.

Backport from 6.0-codebase

6.0-codebase revno: 2617.31.17

----------------------------------------------------------
revno: 2617.22.5
committer: Konstantin Osipov <kostja@sun.com>
branch nick: mysql-6.0-runtime
timestamp: Tue 2009-01-27 05:08:48 +0300
message:
  Remove non-prefixed use of HASH.
  Always use my_hash_init(), my_hash_inited(), my_hash_search(),
  my_hash_element(), my_hash_delete(), my_hash_free() rather
  than non-prefixed counterparts (hash_init(), etc).
  Remove the backward-compatible defines.

----------------------------------------------------------
revno: 2630.28.1
committer: Konstantin Osipov <kostja@sun.com>
branch nick: mysql-6.0-runtime
timestamp: Thu 2008-10-23 15:23:44 +0400
message:
  Bug#32738 "mysqld: Two slow log option/variable descriptions are
  inaccurate"
  Update option description per suggestion from the documentation team.

sql/mysqld.cc:
  Update options help texts (Bug#32738). Use - instead of _ in option
  names.

----------------------------------------------------------
revno: 2630.22.39
committer: Konstantin Osipov <kostja@sun.com>
branch nick: mysql-6.0-runtime
timestamp: Wed 2008-10-08 23:44:34 +0400
message:
  Bug #34481 A typo in HugeTLB error Message.

mysys/my_largepage.c:
  Fix a typo.

Bug #47274 assert in open_table on CREATE TABLE <already existing>

The problem was an assertion during execution of CREATE TABLES. 
This assertion would occur if INSERT DELAYED or REPLACE DELAYED
were used to update a table containing an AUTO_INCREMENT column
and if the inserted row had a user-supplied value for that column.
Any CREATE TABLE statement (including CREATE TABLE SELECT and
CREATE TABLE LIKE) trying to create the same table and 
which followed the INSERT/REPLACED would cause the assertion.

The problem was only noticeable on debug builds of the server
and not present in the mysql-5.1 tree.

The cause of the problem was that the code for delayed insert did
not properly reset the TABLE->auto_increment_if_null flag after 
The flag is used to indicate that a non-null value of an auto_increment field
has been provided by the user or retrieved from a current record.
Open_tables() contains an assertion that tests this flag, and this
was triggered by CREATE TABLE.

This patch fixes the problem by resetting the auto_increment_if_null
field to FALSE once INSERT/REPLACE DELAYED has updated the table, 
similar to what is done already for regular INSERT statements.

Test case added to delayed.test.

----------------------------------------------------------
revno: 2630.22.31
committer: Konstantin Osipov <kostja@sun.com>
branch nick: mysql-6.0-runtime
timestamp: Thu 2008-10-02 19:08:09 +0400
message:
  Bug #34818 --default-table-type option should be removed
  Remove the deprecated option.


sql/mysqld.cc:
  Remove a deprecated option.

----------------------------------------------------------
revno: 2630.22.8
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-runtime
timestamp: Sun 2008-08-10 18:49:52 +0400
message:
  Get rid of typedef struct for the most commonly used types:
  TABLE, TABLE_SHARE, LEX. This simplifies use of tags
  and forward declarations.

revno: 2630.2.16
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-runtime
timestamp: Fri 2008-06-27 13:26:03 +0400
message:
  Fix max_user_connections_func failure on Solaris.
  A connection that failed to log in due to a resource limit could
  be returned to the thread pool with a dangling link to user_connect
  structure of an old user. Later on it could be authenticated
  to a user that doesn't have a resource limit, so this dangling
  link won't be reset. --pool-of-threads mode made the situation
  easy to reproduce, and thus highlighted a bug that has been
  around forever.
  Make sure there are no dangling links.

sql/sql_connect.cc:
  Do not return a connection structure to the thread pool with a dangling link.

----------------------------------------------------------
revno: 2630.7.1
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-lock-tables-new
timestamp: Mon 2008-06-02 15:14:18 +0400
message:
  Fix a test suite timeout in partition.test and partition_csv.test

storage/csv/ha_tina.cc:
  Change the blobroot mem root life cycle and thus fix a memory hog
  in case of many csv handlers.

----------------------------------------------------------
revno: 2630.2.6
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-27430
timestamp: Mon 2008-05-26 16:12:28 +0400
message:
  Cover four special cases of WL#4166 with tests:
  - when the query cache is disabled at the time of prepared statement
  reprepare
  - when long data parameters are used
  - when character_set_connection != character_set_client, and a parameter
  conversion takes place
  - when parameter data is out of acceptable range, e.g. year 10000 is
  supplied as part of MYSQL_TYPE_DATETIME value. The server is supposed
  to warn in such case.

mysql-test/include/query_cache_sql_prepare.inc:
  Addditional test for Bug#27430
mysql-test/r/query_cache_ps_no_prot.result:
  Update result file.
mysql-test/r/query_cache_ps_ps_prot.result:
  Update result file.
tests/mysql_client_test.c:
  Add more tests (Bug#27430 and WL#4166).
  Fix test_datetime_range() test to correctly assert for the number
  of warnings. Additionally, print these warnings out.

-----------------------------------------------------------
revno: 2630.2.4
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-runtime
timestamp: Fri 2008-05-23 02:42:32 +0400
message:
  Bug#27430 "Crash in subquery code when in PS and table DDL changed after
  PREPARE"
  Add a test case for the situation with small TDC and many merge children.

from 6.0-codebase.

mysql-test/r/merge.result:
  Update results (Bug#27430)
mysql-test/t/merge.test:
  Add test case (Bug#27430)

----------------------------------------------------------
revno: 2617.1.12
committer: kostja@bodhi.(none)
timestamp: Sun 2008-04-20 11:18:52 +0400
message:
  A fix for Bug#32771 "events_bugs.test fails randomly".
  In Event_scheduler::stop(), which may be called from destructor,
  wait synchronously for the parallel Event_scheduler::stop() to
  complete before returning. This fixes a race between
  MySQL shutdown thread and the scheduler thread who could call
  stop() in parallel.

sql/event_scheduler.cc:
  There was a race condition between the shutdown thread and
  the scheduler thread: the shutdown thread could delete the scheduler
  mutex before the sheduler has stopped.

local storage for query cache). 
We need more than one pointer in a thread to
represent the query cache and net->query_cache_query can not be used
any more (due to ABI compatibility issues and to different life
time of NET and THD).
This is a backport of the following patch from 6.0:
----------------------------------------------------------
revno: 2476.1157.2
committer: kostja@bodhi.(none)
timestamp: Sat 2007-06-16 13:29:24 +0400


include/mysql.h.pp:
  Update the ABI (NET::query_cache_query is now unused).
include/mysql_com.h:
  NET::query_cache_query is now unused.
sql/mysql_priv.h:
  Update signatures of ex-functios now member functions.
sql/sql_cache.cc:
  Introduce THD::query_cache_tls.
sql/sql_cache.h:
  Introduce THD::query_cache_tls.
sql/sql_class.cc:
  Introduce THD::query_cache_tls.
sql/sql_class.h:
  Introduce THD::query_cache_tls.
sql/sql_parse.cc:
  Use THD::query_cache_tls.
sql/sql_select.cc:
  Use THD::query_cache_tls.

           not on predefined values

The default name of the PID file was constructed, as documented, 
based on the hostname.  This name was subsequently used as the
base for the general log file name.   If the name of the PID
file was overridden in the configuration, and no explicit name
was set for the general log file, the path location for the
PID file was used also for the general log file.
                  
A new variable, 'default_logfile_name', has been introduced.  This name
is constructed based on the hostname, and is then used to
construct both the PID file and the general log file.
                  
The general log file will now, unless explicitly set, be
located in the server data directory (as documentated in
the server docs)


mysql-test/t/log_state.test:
  run/mysqld.log was created as a consequence of this bug.
  After the fix it is no longer created, and will thus not
  be deleted.

Unable to reproduce crash with current version of the 5.5.0 codebase.
Test case for MyISAM/InnoDB based on the bug rapport added to 
sp_trans.test.

Backport of revno: 2617.65.9.

     
Unable to reproduce error on current version of the 5.5.0
codebase. Test case based on the bug report added to trigger.test.

Backport of revno: 2617.52.11.

Original revision in 6.0:
------------------------------------------------------------
revno: 2630.13.14
committer: Alexander Nozdrin <alik@mysql.com>
branch nick: 6.0-rt-fix
timestamp: Fri 2008-07-25 22:44:20 +0400
message:
  Fix test & result files (WL 4300).
------------------------------------------------------------

Original revision in 6.0:
------------------------------------------------------------
revno: 2630.13.11
committer: Alexander Nozdrin <alik@mysql.com>
branch nick: 6.0-rt-wl4300
timestamp: Thu 2008-07-24 11:44:21 +0400
message:
  A patch for WL#4300: Define privileges for tablespaces.
------------------------------------------------------------

per-file messages:
  mysql-test/r/grant.result
    Update result file: new columm 'Create_tablespace_priv' has been added to mysql.user.
  mysql-test/r/ps.result
    Update result file: new columm 'Create_tablespace_priv' has been added to mysql.user.
  mysql-test/r/system_mysql_db.result
    Update result file: new columm 'Create_tablespace_priv' has been added to mysql.user.
  mysql-test/suite/falcon/r/falcon_tablespace_priv.result
    Test case for WL#4300.
  mysql-test/suite/falcon/t/falcon_tablespace_priv.test
    Test case for WL#4300.
  mysql-test/suite/ndb/r/ndb_dd_ddl.result
    Test case for WL#4300.
  mysql-test/suite/ndb/t/ndb_dd_ddl.test
    Test case for WL#4300.
  scripts/mysql_system_tables.sql
    New columm 'Create_tablespace_priv' has been added to mysql.user.
  scripts/mysql_system_tables_data.sql
    'CREATE TABLESPACE' is granted by default to the root user.
  scripts/mysql_system_tables_fix.sql
    Grant 'CREATE TABLESPACE' privilege during system table upgrade
    if a user had SUPER privilege.
  sql/sql_acl.cc
    Added CREATE TABLESPACE privilege.
  sql/sql_acl.h
    Added CREATE TABLESPACE privilege.
  sql/sql_parse.cc
    Check global 'CREATE TABLESPACE' privilege for the following SQL statements:
      - CREATE | ALTER | DROP TABLESPACE
      - CREATE | ALTER | DROP LOGFILE GROUP
  sql/sql_show.cc
    Added CREATE TABLESPACE privilege.
  sql/sql_yacc.yy
    Added CREATE TABLESPACE privilege.

           strict SQL mode

The problem was that a COMMENT longer than 64 characters
caused CREATE PROCEDURE to fail.

This patch fixed the problem by changing the COMMENT field in 
mysql.proc from char(64) to text. The corresponding ROUTINE_COMMENT
field in INFORMATION_SCHEMA.ROUTINES is also changed from
varchar(64) to longtext.

mysql_system_tables.sql and mysql_system_tables_fix.sql updated.
Test case added to sp.test and affected result-files updated.

case than in corr index".

Server was unable to find existing or explicitly created supporting
index for foreign key if corresponding statement clause used field
names in case different than one used in key specification and created
yet another supporting index.
In cases when name of constraint (and thus name of generated index)
was the same as name of existing/explicitly created index this led
to duplicate key name error.

The problem was that unlike all other code Key_part_spec::operator==()
compared field names in case sensitive fashion. As result routines
responsible for getting rid of redundant generated supporting indexes
for foreign key were not working properly for versions of field names
using different cases.

mysql-test/r/innodb_mysql.result:
  Added test case for bug #39932 "create table fails if column for FK
  is in different case than in corr index".
mysql-test/t/innodb_mysql.test:
  Added test case for bug #39932 "create table fails if column for FK
  is in different case than in corr index".
sql/sql_class.cc:
  Make field name comparison case-insensitive like it is
  in the rest of server.

Original revision in 6.0:
------------------------------------------------------------
revno: 2599.108.1
committer: Alexander Nozdrin <alik@sun.com>
branch nick: 6.0-rpl-bug40128
timestamp: Wed 2009-01-21 15:33:42 +0300
message:
  Fix for Bug#40128: drop-no_root fails under windows in 6.0-rpl.
  
  The problem was that directories with no permission (000) files
  are deleted differently on UNIX and on Windows.
  
  On UNIX, 000-permission file is deleted perfectly, but other files
  are left in the directory.
  
  On Windows, 000-permission file is not deleted, but other files
  are deleted.
  
  Also, the fix needed a change in mysqltest.c: 'chmod' directive
  should return a positive error code (in order to be handled).
  It's decided to return a constant '1' for all error codes
  just to be OS-independent.
------------------------------------------------------------

           InnoDB specified.
            
NO_ENGINE_SUBSTITUTION added to TRADITIONAL sql mode to prevent
silent conversions from InnoDB to MyISAM in that sql mode.

A number of test case results files updated to reflect this change.
Test added to sql_mode.test that checks that TRADITIONAL really 
includes NO_ENGINE_SUBSTITUION.


mysql-test/t/ctype_utf8.test:
  This test lacked "--source include/have_innodb.inc" which meant that 
  a number of DDL statements with engine=innodb in reality were using 
  myisam. "--disable_warnings" around all such statements, meant that 
  these engine substitutions were not visible. Test case has been 
  updated to include have_innodb.inc and "--disable_warnings" have been 
  removed for the relevant DDL statements.

storing and restoring information about foreign keys in the .FRM files and
properly displaying it in SHOW CREATE TABLE output and I_S tables.

The idea of this patch is to change type of Key_part_spec::field_name and
Key::name to LEX_STRING in order to avoid extra strlen() calls during
semantic analysis and statement execution, particularly, in code to be
implemented on the 2nd milestone of WL#148.

Note that since we are not using LEX_STRING everywhere yet (e.g. in
Create_field and KEY) and we want to limit scope of our changes we
have to do strlen() in places where we create Key and Key_part_spec
instances from objects using plain (char*) for strings. These calls
will go away during the process of further (char*) -> LEX_STRING
refactoring.

We have introduced these changes in 6.0 and backported them to 5.5
tree to make people aware of these changes as early as possible and
to simplify merges with mysql-fk and mysql-6.1-fk trees.

No test case is needed since this patch does not introduce any
user visible changes.

sql/sql_class.cc:
  Key_part_spec::field_name is now LEX_STRING. Adjusted code accordingly.
sql/sql_class.h:
  Changed type of Key_part_spec::field_name and Key::name to LEX_STRING in
  order to avoid extra strlen() calls in code responsible for semantic
  analysis and statement execution (e.g. in future code responsible for
  saving/restoring info about foreign keys).
sql/sql_lex.cc:
  Moved null_lex_str from sql_yacc.yy to sql_lex.cc and added its
  declaration to sql_lex.h to make it accessible in other SQL-layer
  modules (e.g. sql_parse.cc).
sql/sql_lex.h:
  Made null_lex_str accessible from outside of sql_lex.cc.
sql/sql_parse.cc:
  Key_part_spec::field_name and Key::name are now LEX_STRING. Adjusted
  code accordingly.
sql/sql_table.cc:
  Adjusted code to accomodate change of type to LEX_STRING for
  Key_part_spec::field_name and Key::name.
sql/sql_yacc.yy:
  Now Key::name and Key_part_spec::field_name are LEX_STRINGs. Adjusted
  grammar to be able properly initialize them. This should allow us to
  save on some strlen() calls during later stages of statement execution.

           "have_profiling"

1) Renamed have_community_features server system variable to
have_profiling.
2) Removed eable-community-features configure option and
ENABLE_COMMUNITY_FEATURES macro.
3) Removed COMMUNITY_SERVER macro and replaced its usage by 
ENABLED_PROFILING.

Only --enable-profiling is now needed to enable profiling.
It was the only existing "community feature", so there was
no need for both configure options.

Using --enable-community-features will give a warning message
since it no longer exists.

To-number conversion warnings work differenly with CHAR 
and VARCHAR sp variables.

The original revision-IDs are:
  staale.smedseng@sun.com-20081124095339-2qdvzkp0rn1ljs30
  staale.smedseng@sun.com-20081125104611-rtxic5d12e83ag2o
                                                
The patch provides ER_TRUNCATED_WRONG_VALUE warning messages
for conversion of VARCHAR to numberic values, in line with
messages provided for CHAR conversions. Conversions are
checked for success, and the message is emitted in case
failure.
                                                
The tests are amended to accept the added warning messages,
and explicit conversion of ON/OFF values is added for
statements checking system variables. In test
rpl.rpl_switch_stm_row_mixed checking for warnings is
temporarily disabled for one statement, as this generates
warning messages for strings that vary between executions.


sql/field.cc:
  The pushing of the truncation warning is now done in a
  separate static function, and used in various places.

An ALTER TABLE statement which added a column and added
a non-partial index on it failed with:
            
"ERROR 1089 (HY000): Incorrect sub part key; the used
key part isn't a string, the used length is longer than
the key part, or the storage engine doesn't support unique
sub keys"
            
In a check introduced to fix an earlier bug (no. 26794),
to allow for indices on spatial type columns, the
test expression was flawed (a logical OR was used instead
of a logical AND), which led to this regression.
            
The code in question does a sanity check on the key, and
the flawed code mistakenly classified any index created
in the way specified above as a partial index.  Since
many data types does not allow partial indices, the
statement would fail.