cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server
Crash while fetching from table with 5 million records."
A fix for a possible memory leak when fetching into an SP cursor
in a long loop.
The patch uses a common implementation of cursors in the binary protocol and
in stored procedures and implements materialized cursors.
For implementation details, see comments in sql_cursor.cc
include/my_sys.h:
- declaration for multi_alloc_root
libmysqld/Makefile.am:
- drop protocol_cursor.cc, add sql_cursor.cc (replaces the old
implementation of cursors with a new one)
mysql-test/r/ctype_ujis.result:
- test results fixed (a test case for Bug#6513)
mysql-test/r/sp-big.result:
- test results fixed (a test case for Bug#9819)
mysql-test/t/ctype_ujis.test:
Add a test case for Bug#6513 "Test Suite: Values inserted by using cursor is
interpreted latin1 character"
mysql-test/t/sp-big.test:
Add a restricted test case for Bug#9819 "Cursors: Mysql Server Crash
while fetching from table with 5 million records."
mysys/my_alloc.c:
- an implementation of multi_alloc_root; this is largely a copy-paste
from mulalloc.c, but the function is small and there is no easy way
to reuse the existing C function.
sql/Makefile.am:
- add sql_cursor.h, sql_cursor.cc (a new implementation of stored procedure
cursors) and drop protocol_cursor.cc (the old one)
sql/handler.cc:
- now TABLE object has its mem_root always initialized.
Adjust the implementation handler::ha_open
sql/item_subselect.cc:
- adjust to the changed declaration of st_select_lex_unit::prepare
sql/protocol.h:
- drop Protocol_cursor
sql/sp_head.cc:
- move juggling with Query_arena::free_list and Item::next to
sp_eval_func_item, as this is needed in 3 places already.
sql/sp_head.h:
- declare a no-op implementation for cleanup_stmt in sp_instr_cpush.
This method is needed for non-materializing cursors, which are yet not
used in stored procedures.
- declaration for sp_eval_func_item
sql/sp_rcontext.cc:
- reimplement sp_cursor using the new implementation of server side cursors.
- use sp_eval_func_item to assign values of SP variables from the
row fetched from a cursor. This should fix a possible memory leak in
the old implementation of sp_cursor::fetch
sql/sp_rcontext.h:
- reimplement sp_cursor using the new implementation of server side cursors.
sql/sql_class.cc:
- disable the functionality that closes transient cursors at commit/rollback;
transient cursors are not used in 5.0, instead we use materialized ones.
To be enabled in a later version.
sql/sql_class.h:
- adjust to the rename Cursor -> Server_side_cursor
- additional declarations of select_union used in materialized cursors
sql/sql_derived.cc:
- reuse bits of tmp table code in UNION, derived tables, and materialized
cursors
- cleanup comments
sql/sql_lex.h:
- declarations of auxiliary methods used by materialized cursors
- a cleanup in st_select_lex_unit interface
sql/sql_list.h:
- add an array operator new[] to class Sql_alloc
sql/sql_prepare.cc:
- split the tight coupling of cursors and prepared statements to reuse
the same implementation in stored procedures
- cleanups of error processing in Prepared_statement::{prepare,execute}
sql/sql_select.cc:
- move the implementation of sensitive (non-materializing) cursors to
sql_cursor.cc
- make temporary tables self-contained: the table, its record and fields
are allocated in TABLE::mem_root. This implementation is not clean
and resets thd->mem_root several times because of the way create_tmp_table
works (many additional things are done inside it).
- adjust to the changed declaration of st_select_lex_unit::prepare
sql/sql_select.h:
- move the declaration of sensitive (non-materializing) cursors to
sql_cursor.cc
sql/sql_union.cc:
- move pieces of st_select_unit::prepare to select_union and st_table
methods to be able to reuse code in the implementation of materialized
cursors
sql/sql_view.cc:
- adjust to the changed signature of st_select_lex_unit::prepare
sql/table.cc:
- implement auxiliary st_table methods for use with temporary tables
sql/table.h:
- add declarations for auxiliary methods of st_table used to work with
temporary tables
tests/mysql_client_test.c:
- if cursors are materialized, a parallel update of the table used
in the cursor may go through: update the test.
sql/sql_cursor.cc:
New BitKeeper file ``sql/sql_cursor.cc'' -- implementation of server side
cursors
sql/sql_cursor.h:
New BitKeeper file ``sql/sql_cursor.h'' - declarations for
server side cursors.
thd->allow_sum_func was left 'true' after previous statement thus allowing
sum funcs to be present in conditions.
thd->allow_sum_func should be set to 0 for each query and each prepared
statement reinitialization. This is done in lex_start() and
reset_stmt_for_execute().
sql/sql_lex.cc:
Fix bug#13180 thd->allow_sum_func wasn't reset obefore query processing.
thd->allow_sum_func is set to 0 in lex_start().
sql/sql_prepare.cc:
Fix bug#13180 thd->allow_sum_func wasn't reset obefore query processing.
thd->allow_sum_func is set to 0 in reset_stmt_for_execute().
mysql-test/t/update.test:
Test case for bug#13180 thd->allow_sum_funcs wasn't reset before query
processing.
mysql-test/r/update.result:
Test case for bug#13180 thd->allow_sum_funcs wasn't reset before query
processing.
into serg.mylan:/usr/home/serg/Abk/mysql-5.0-merged
sql/ha_myisam.cc:
Auto merged
sql/handler.h:
Auto merged
sql/mysql_priv.h:
Auto merged
sql/mysqld.cc:
Auto merged
sql/set_var.cc:
Auto merged
sql/slave.cc:
Auto merged
sql/sql_class.h:
Auto merged
sql/sql_parse.cc:
Auto merged
sql/sql_yacc.yy:
Auto merged
into perch.ndb.mysql.com:/home/jonas/src/mysql-5.0
ndb/src/kernel/blocks/backup/Backup.cpp:
Auto merged
ndb/src/kernel/blocks/dbtup/Dbtup.hpp:
Auto merged
ndb/src/kernel/blocks/dbtup/DbtupTrigger.cpp:
merge
ndb backup with charsets and simultanious updates
make sure that backup _dont_ read xfrm data
ndb/src/kernel/blocks/backup/Backup.cpp:
Backup can do dirty read
ndb/src/kernel/blocks/dbtup/Dbtup.hpp:
Fix so that backup dont read xfrm data
ndb/src/kernel/blocks/dbtup/DbtupTrigger.cpp:
Fix so that backup dont read xfrm data
VC++Files/mysys/mysys.dsp:
Auto merged
VC++Files/mysys/mysys_ia64.dsp:
Auto merged
client/mysql.cc:
Auto merged
include/my_sys.h:
Auto merged
myisam/mi_check.c:
Auto merged
myisam/myisamchk.c:
Auto merged
mysql-test/mysql-test-run.pl:
Auto merged
mysql-test/r/rpl_multi_delete2.result:
Auto merged
mysql-test/t/innodb.test:
Auto merged
mysql-test/t/rpl_multi_delete2-slave.opt:
Auto merged
mysql-test/t/rpl_multi_delete2.test:
Auto merged
sql/ha_myisam.cc:
Auto merged
sql/opt_sum.cc:
Auto merged
sql/set_var.cc:
Auto merged
sql/sql_class.h:
Auto merged
sql/sql_parse.cc:
Auto merged
mysql-test/r/union.result:
a fix (bug #7589: Decimal types are ignored when updating data from another column).
removed wrong test result (see #13372)
mysql-test/t/union.test:
a fix (bug #7589: Decimal types are ignored when updating data from another column).
removed wrong test result (see #13372)
sql/field_conv.cc:
a fix (bug #7589: Decimal types are ignored when updating data from another column).
use memcpy() only for identical decimals.
mysql-test/mysql-test-run.pl:
Not all args to mysqltest should be added to the MYSQL_TEST environment variable. The MYSQL_TEST var is used when testing mysqltest and --record and result_file should not be set.
The problem was in that the MIN/MAX optimization in opt_sum_query was
replacing MIN/MAX functions with their constant argument without
taking into account that a query has no result rows.
mysql-test/r/func_group.result:
Test for BUG#12882.
mysql-test/t/func_group.test:
Test for BUG#12882.
sql/item_sum.cc:
If it is known that a query has no result rows, do not call add()
via the call to Item_sum::no_rows_in_result() which calls reset().
Instead directly call clear() so that the MIN and MAX functions
produce NULL when there are no result rows.
sql/opt_sum.cc:
* Do not apply MIN/MAX optimization when the operand of MIN/MAX is
a constant if it can't be determined whether the query has any
result rows. The reason is that if the query has result rows,
then the result of MIN/MAX is its constant argument, but if the
query result is empty, then the result of MIN/MAX must be NULL
irrespective of its argument.
* The patch also simplifies a bit the branch that hadles COUNT().
libmysqld/lib_sql.cc:
struct st_security_context nad to class Security_context
sql/item_func.cc:
fixed method call, name and contence to be compatible with new code
sql/item_func.h:
fixed method to be compatible wit new code
sql/sql_parse.cc:
fixed typo
removed compiler warnings
statistics (like 4.0 did) (patch #3, with review #1 & #2 feedback addressed)
include/myisam.h:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
myisam/mi_check.c:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
myisam/myisamchk.c:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
mysql-test/r/myisam.result:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
mysql-test/t/myisam.test:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
sql/ha_myisam.cc:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
sql/handler.h:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
sql/mysqld.cc:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
sql/set_var.cc:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
sql/sql_class.h:
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics
into sanja.is.com.ua:/home/bell/mysql/bk/work-owner5-5.0
sql/ha_innodb.cc:
Auto merged
sql/item_strfunc.cc:
Auto merged
sql/sql_base.cc:
Auto merged
sql/sql_show.cc:
Auto merged
sql/sql_view.cc:
Auto merged
sql/item_func.cc:
merge
Part 2 postreview fixes.
sql/ha_innodb.cc:
Renamed structure st_security_context to class Security_context
sql/item_func.cc:
Renamed structure st_security_context to class Security_context
sql/item_strfunc.cc:
fixed USER() function
sql/log.cc:
variable used to optimize access to security context
sql/mysql_priv.h:
Renamed structure st_security_context to class Security_context
sql/mysqld.cc:
main security context used direcly
sql/sp_head.cc:
Renamed structure st_security_context to class Security_context
removed unneed variable
sql/sp_head.h:
Comment inmroved
Renamed structure st_security_context to class Security_context
sql/sql_acl.cc:
Renamed structure st_security_context to class Security_context
fixed function comment and return value
variable used to optimize access to security context
Renamed method of Security_ontext
sql/sql_acl.h:
fixed return value type
sql/sql_class.cc:
Renamed structure st_security_context to class Security_context
sql/sql_class.h:
Renamed structure st_security_context to class Security_context
Method renamed
sql/sql_db.cc:
Renamed structure st_security_context to class Security_context
fixed layout
sql/sql_parse.cc:
registration of wanted access for underlying tables
sql/sql_show.cc:
Renamed structure st_security_context to class Security_context
fixed layout
sql/sql_yacc.yy:
Renamed structure st_security_context to class Security_context
Not fixed in 4.1 as not critical. Also I'm correcting error checking of multi-UPDATE/DELETE
when it comes to binlogging, to make it consistent with when we rollback the statement.
mysql-test/r/rpl_multi_delete.result:
result update
mysql-test/r/rpl_multi_update.result:
result update
mysql-test/t/rpl_multi_delete.test:
test for BUG#13348
mysql-test/t/rpl_multi_update.test:
test of BUG#13348
sql/sql_delete.cc:
We now binlog multi-DELETE even if no row was updated (like we do for DELETE).
I'm also correcting some error checking (< instead of <=), basing myself on when we rollback.
sql/sql_update.cc:
we now binlog multi-UPDATE even if no row was updated (like we do for UPDATE).
Adding to existing tests to test new behaviour.
I'm also correcting some error checking (< instead of <=), basing myself on when we rollback.
The problem was in the way table references are pre-filtered when
resolving a qualified field. When resolving qualified table references
we search recursively in the operands of the join. If there is
natural/using join with a merge view, the first call to find_field_in_table_ref
makes a recursive call to itself with the view as the new table reference
to search for the column. However the view has both nested_join and
join_columns != NULL so it skipped the test whether the view name matches
the field qualifier. As a result the field was found in the view since the
view already has a field with the same name. Thus the field was incorrectly
resolved as the view field.
mysql-test/r/select.result:
Test for BUG#13127.
mysql-test/t/select.test:
Test for BUG#13127.
sql/sql_base.cc:
The patch contains two independent changes:
- When searching for qualified fields, include merge views and information schema tables
used with SHOW (both using TABLE_LIST::field_translation to represent result fields) in
the test that compares the qualifying table with the name of the table reference being
searched. This change fixes the bug.
- Do not search the materialized list of columns of a NATURAL/USING join if 'table_list'
is a stored table or merge view. Instead search directly in the table or view as if it
is not under a natural join. This is a performance improvement since if 'table_list'
is a stored table, then the search can utilize the name hash for table names.
