ONLY_FULL_GROUP_BY
The check for non-aggregated columns in queries with aggregate function, but without
GROUP BY was treating all the parts of the query as if they are in the SELECT list.
Fixed by ignoring the non-aggregated fields in the WHERE clause.
returns truncated results
Problem: performig conversion from {INT, DECIMAL, REAL} to CHAR
we incorrectly set its max length in some cases that may lead
to truncated results returned.
Fix: properly set CONVERT({INT, DECIMAL, REAL}, CHAR) result's
max length.
It was possible to crash a mysqld build with EXTRA_DEBUG using
CREATE TABLE ... COMMENT with a specially-crafted UTF-8 string.
This CS removes the check that caused it since it no longer
applies in current servers anyway, and adds comments instead
to avoid future confusion.
Problem: memory leak occurs when we open a federated table
that has its share in the hash.
Fix: free not used memory.
Note: the fix should NOT be merged to 5.1 (the code changed).
When a CSV file contained comma separated elements
that were not enclosed in quotes, it was causing the
mysql server to crash.
The old algorithm that parsed the content of a row in
mysql 5.0 was assuming that the values of the fields
in a .CSV file will be enclosed in quotes and will be
separated by commas.
This was causing the old algorithm to fail when the
content of the file resembled the following
3,"sans quotes"
The CSV engine that is part of mysql 5.0 was expecting
the above to be
"3","sans quotes"
The above is just one example of where the engine was
failing for what would be recognized as a valid .CSV
file content otherwise.
The proposed fix changes the old algorithm being used
to parse rows from the .CSV file to handle two separate
cases
1) When the current field of the row is enclosed in quotes
2) When the current field of the row is not enclosed in
quotes
Item_func_div didn't calculate the precision of the result properly.
The result of 5/0.0001 is 5000 so we have to add decimals of the divisor
to the planned precision.
per-file comments:
mysql-test/r/type_newdecimal.result
Bug#31616 div_precision_increment description looks wrong
test result fixed
mysql-test/t/type_newdecimal.test
Bug#31616 div_precision_increment description looks wrong
test case
sql/item_func.cc
Bug#31616 div_precision_increment description looks wrong
precision must be increased with args[1]->decimals parameter
missing after downgrade
Obsolete arc/ directory and view .frm file backup support
has been removed by the patch for bug 17823. However, that
bugfix caused a problem with "live downgrades" of the
server: if we rename some view 4 times under 5.1.29/5.0.72
and then try to rename it under 5.1.28/5.0.70 on the same
database, the server fails with a error:
query 'RENAME TABLE ... TO ...' failed: 6: Error on
delete of '....frm-0001' (Errcode: 2)
Also .frm file of that view may be lost (renamed to .frm~).
The server failed because it tried to rename latest 3
backup .frm files renaming the view: the server used an
integer value of the "revision" field of .frm file to
extract those file names. After the fix for bug 17823 those
files were not created/maintained any more, however the
"revision" field was incremented as usual. So, the server
failed renaming non existent files.
This fix solves the problem by removing the support for
"revision" .frm file field:
1. New server silently ignores existent "revision" fields
in old .frm files and never write it down;
2. Old server assumes, that missing "revision" field in new
.frm files means default value of 0.
3. Accordingly to the fix for bug 17823 the new server
drops arc/ directory on alter/rename view, so after
"live downgrade" old server begins maintenance of the
arc/ directory from scratch without conflicts with .frm
files.
status
The problem appears to be a race condition, when service is being
stopped right after startup. We set the service status to SERVICE_RUNNING
way too early it cannot yet handle stop requests - initialization has
not finished and hEventShutdown that signals server to stop is not yet
created. If somebody issues "net stop MySQL" at this time, MySQL is not
informed about the stop and continues to run as usual, while
NTService::ServiceMain() stucks forever waiting for mysql's "main" thread
to finish.
Solution is to remain in SERVICE_START_PENDING status until after server
initialization is fully complete and only then change the status to
SERVICE_RUNNING. In SERVICE_START_PENDING we do not accept service control
requests, i.e it is not possible to stop service in that time.
With fix for bug 25951 index hints are ignored for fulltext
searches, as handling of fulltext indexes is different from
handling regular indexes. Meaning it is not possible to
implement true index hints support for fulltext indexes within
the scope of current fulltext architecture.
The problem is that prior to fix for bug 25951, some useful
index hints still could be given for boolean mode searches.
This patch implements special index hints support for fulltext
indexes with the following characteristics:
- all index hints are still ignored for NLQ mode searches -
it cannot work without an index;
- for 5.1 and up index hints FOR ORDER BY and FOR GROUP BY are
still ignored for fulltext indexes;
- boolean mode searches honor USE/FORCE/IGNORE INDEX hints;
- as opposed to index hints for regular indexes, index hints
for fulltext BOOLEAN mode searches affect the usage of the
index for the whole query.
A string buffers which were included in the 'view' data structure
were allocated on the stack, causing an invalid pointer when used
after the function returned.
The fix: use copy of values for view->md5 & view->queries
- Make send_row_on_empty_set() return FALSE when simplify_cond() has found out
that HAVING is always FALSE
re-committing to put the fix into 5.0 and 5.1
Visual Studio 2008 Express edition does not include message compiler mc.exe
It is not possible to build MySQL server if only VC2008 Express is installed,
because we use mc.exe to generate event log messages.
This patch removes the mc.exe dependency. Generated files message.h,
message.rc and MSG00001.bin are checked into source code repository.
Instructions on how to add or change messages are added to messages.mc
The problem was that the server did not robustly handle a
unilateral roll back issued by the Resource Manager (RM)
due to a resource deadlock within the transaction branch.
By not acknowledging the roll back, the server (TM) would
eventually corrupt the XA transaction state and crash.
The solution is to mark the transaction as rollback-only
if the RM indicates that it rolled back its branch of the
transaction.
fails after the first time
Two separate problems :
1. When flattening joins the linked list used for name resolution
(next_name_resolution_table) was not updated.
Fixed by updating the pointers when extending the table list
2. The items created by expanding a * (star) as a column reference
were marked as fixed, but no cached table was assigned to them
(unlike what Item_field::fix_fields does).
Fixed by assigning a cached table (so the re-preparation is done
faster).
Note that the fix for #2 hides the fix for #1 in most cases
(except when a table reference cannot be cached).
Server crashed during a sort order optimization
of a dependent subquery:
SELECT
(SELECT t1.a FROM t1, t2
WHERE t1.a = t2.b AND t2.a = t3.c
ORDER BY t1.a)
FROM t3;
Bitmap of tables, that the reference to outer table
column uses, in addition to the regular table bit
has the OUTER_REF_TABLE_BIT bit set.
The only_eq_ref_tables function traverses this map
bit by bit simultaneously with join->map2table list.
Obviously join->map2table never contains an entry
for the OUTER_REF_TABLE_BIT pseudo-table, so the
server crashed there.
The only_eq_ref_tables function has been modified
to traverse regular table bits only like the
update_depend_map function (resetting of the
OUTER_REF_TABLE_BIT there is enough, but
resetting of the whole set of PSEUDO_TABLE_BITS
is used there for sure).
The problem is that the offset argument of the limit clause
might be truncated on a 32-bits server built without big
tables support. The truncation was happening because the
original 64-bits long argument was being cast to a 32-bits
(ha_rows) offset counter.
The solution is to check if the conversing resulted in value
truncation and if so, the offset is set to the maximum possible
value that can fit on the type.
If delayed insert fails to upgrade the lock it was not
freeing the temporary memory storage used to keep
newly constructed blob values in memory.
Fixed by iterating over the remaining rows in the delayed
insert rowset and freeing the blob storage for each row.
No test suite because it involves concurrent delayed inserts
on a table and cannot easily be made deterministic.
Added a correct valgrind suppression for Fedora 9.
The problem is that field names constructed due to wild-card
expansion done inside a stored procedure could point to freed
memory if the expansion was performed after the first call to
the stored procedure.
The problem was solved by patch for Bug#38691. The solution
was to allocate the database, table and field names in the
in the statement memory instead of table memory.
Select with a "NULL NOT IN" condition containing complex
subselect from the same table as in the outer select failed
with an assertion.
The failure was caused by a concatenation of circumstances:
1) an inner select was optimized by make_join_statistics to use
the QUICK_RANGE_SELECT access method (that implies an index
scan of the table);
2) a subselect was independent (constant) from the outer select;
3) a condition was pushed down into inner select.
During the evaluation of a constant IN expression an optimizer
temporary changed the access method from index scan to table
scan, but an engine handler was already initialized for index
access by make_join_statistics. That caused an assertion.
Unnecessary index initialization has been removed from
the QUICK_RANGE_SELECT::init method (QUICK_RANGE_SELECT::reset
reinvokes this initialization).
with COALESCE and JOIN
The server returned to a client the VARBINARY column type
instead of the DATE type for a result of the COALESCE,
IFNULL, IF, CASE, GREATEST or LEAST functions if that result
was filesorted in an anonymous temporary table during
the query execution.
For example:
SELECT COALESCE(t1.date1, t2.date2) AS result
FROM t1 JOIN t2 ON t1.id = t2.id ORDER BY result;
To create a column of various date/time types in a
temporary table the create_tmp_field_from_item() function
uses the Item::tmp_table_field_from_field_type() method
call. However, fields of the MYSQL_TYPE_NEWDATE type were
missed there, and the VARBINARY columns were created
by default.
Necessary condition has been added.
derived table cause crash
When a multi-UPDATE command fails to lock some table, and
subsequently succeeds, the tables need to be reopened if
they were altered. But the reopening procedure failed for
derived tables.
Extra cleanup has been added.
When running Stored Routines the Status Variable "Questions" was wrongly
incremented. According to the manual it should contain the "number of
statements that clients have sent to the server"
Introduced a new status variable 'questions' to replace the query_id
variable which currently corresponds badly with the number of statements
sent by the client.
The new behavior is ment to be backward compatible with 4.0 and at the
same time work with new features in a similar way.
This is a backport from 6.0
``FLUSH TABLES WITH READ LOCK''
Concurrent execution of 1) multitable update with a
NATURAL/USING join and 2) a such query as "FLUSH TABLES
WITH READ LOCK" or "ALTER TABLE" of updating table led
to a server crash.
The mysql_multi_update_prepare() function call is optimized
to lock updating tables only, so it postpones locking to
the last, and if locking fails, it does cleanup of modified
syntax structures and repeats a query analysis. However,
that cleanup procedure was incomplete for NATURAL/USING join
syntax data: 1) some Field_item items pointed into freed
table structures, and 2) the TABLE_LIST::join_columns fields
was not reset.
Major change:
short-living Field *Natural_join_column::table_field has
been replaced with long-living Item*.
The '@' symbol can not be used in the host name according to rfc952.
The fix:
added function check_host_name(LEX_STRING *str)
which checks that all symbols in host name string are valid and
host name length is not more than max host name length
(just moved check_string_length() function from the parser into check_host_name()).
The problem:
I_S views table does not check the presence of SHOW_VIEW_ACL|SELECT_ACL
privileges for a view. It leads to discrepancy between SHOW CREATE VIEW
and I_S.VIEWS.
The fix:
added appropriate check.
When analyzing the possible index use cases the server was re-using an internal structure.
This is wrong, as this internal structure gets updated during the analysis.
Fixed by making a copy of the internal structure for every place it needs to be used.
Also stopped the generation of empty SEL_TREE structures that unnecessary
complicate the analysis.
from stored procedure.
Problem: we replace all references to local variables in stored procedures
with NAME_CONST(name, value) logging to the binary log. However, if the
value's collation differs we might get an 'illegal mix of collation'
error as we don't pass the collation to the function.
Fix: pass the value's collation to NAME_CONST().
Note: actually we should pass to NAME_CONST() the value's derivation as well.
It's impossible without the parser modifying. Now we always set the
derivation to DERIVATION_IMPLICIT, the same as local variables have.
Server created "arc" directories inside database directories and
maintained there useless copies of .frm files.
Creation and renaming procedures of those copies as well as
creation of "arc" directories has been discontinued.
Removal procedure has been kept untouched to be able to
cleanup existent database directories by the DROP DATABASE
query. Also view renaming procedure has been updated to remove
these directories.
JOIN for the subselect wasn't cleaned if we came upon an error
during sub_select() execution. That leads to the assertion failure
in close_thread_tables()
part of the 6.0 code backported
per-file comments:
mysql-test/r/sp-error.result
Bug#37949 Crash if argument to SP is a subquery that returns more than one row
test result
mysql-test/t/sp-error.test
Bug#37949 Crash if argument to SP is a subquery that returns more than one row
test case
sql/sp_head.cc
Bug#37949 Crash if argument to SP is a subquery that returns more than one row
lex->unit.cleanup() call added if not substatement
A stored procedure involving substrings could crash the server on certain
platforms because of invalid memory reads.
During storing the new blob-field value, the cached value's address range
overlapped that of the new field value. This caused problems when the
cached value storage was reallocated to provide access for a new
characater set representation. The patch checks the address ranges, and if
they overlap, the new field value is copied to a new storage before it is
converted to the new character set.
The fix for bug 31887 was incomplete : it assumes that all the
field types returned by the IS_NUM macro are descendants of
Item_num and tries to zero-fill the values before doing constant
substitution with such fields when they are compared to constant string
values.
The only exception to this is Field_timestamp : it's in the IS_NUM
macro, but is not a descendant of Field_num.
Fixed by excluding timestamp fields (Field_timestamp) when zero-filling
when converting the constant to compare with to a string.
Note that this will not exclude the timestamp columns from const
propagation.
The problem here is that symbols can not be loaded, because symbol
path is not set and default path does not include the directory
where PDB is located.
The problem is _not_ reproducible on the same machine where
mysqld.exe is built - if PDB is not found in the symbol path,
dbghelp would fallback to fully qualified PDB path as given in the
executable header and on the build host this will succeed.
The solution is to calculate symbol path and pass it to SymInitialize()
call.
in open_table()
Problem: repeating "CREATE... ( AUTOINCREMENT) ... SELECT" may lead to
an assertion failure.
Fix: reset table->auto_increment_field_not_null after each record
writing.
INSERT .. SELECT .. ON DUPLICATE KEY UPDATE col=DEFAULT
In order to get correct values from update fields that
belongs to the SELECT part in the INSERT .. SELECT .. ON
DUPLICATE KEY UPDATE statement, the server adds referenced
fields to the select list. Part of the code that does this
transformation is shared between implementations of
the DEFAULT(col) function and the DEFAULT keyword (in
the col=DEFAULT expression), and an implementation of
the DEFAULT keyword is incomplete.
returns unexpected result
If:
1. a table has a not nullable BIT column c1 with a length
shorter than 8 bits and some additional not nullable
columns c2 etc, and
2. the WHERE clause is like: (c1 = constant) AND c2 ...,
the SELECT query returns unexpected result set.
The server stores BIT columns in a tricky way to save disk
space: if column's bit length is not divisible by 8, the
server places reminder bits among the null bits at the start
of a record. The rest bytes are stored in the record itself,
and Field::ptr points to these rest bytes.
However if a bit length of the whole column is less than 8,
there are no remaining bytes, and there is nothing to store in
the record at its regular place. In this case Field::ptr points
to bytes actually occupied by the next column in a record.
If both columns (BIT and the next column) are NOT NULL,
the Field::eq function incorrectly deduces that this is the
same column, so query transformation/equal item elimination
code (see build_equal_items_for_cond) may mix these columns
and damage conditions containing references to them.
used causes server crash.
When the loose index scan access method is used values of aggregated functions
are precomputed by it. Aggregation of such functions shouldn't be performed
in this case and functions should be treated as normal ones.
The create_tmp_table function wasn't taking this into account and this led to
a crash if a query has MIN/MAX aggregate functions and employs temporary table
and loose index scan.
Now the JOIN::exec and the create_tmp_table functions treat MIN/MAX aggregate
functions as normal ones when the loose index scan is used.
test_if_data_home_dir fixed to look into real path.
Checks added to mi_open for symlinks into data home directory.
per-file messages:
include/my_sys.h
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
my_is_symlink interface added
include/myisam.h
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
myisam_test_invalid_symlink interface added
myisam/mi_check.c
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
mi_open_datafile calls modified
myisam/mi_open.c
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
code added to mi_open to check for symlinks into data home directory.
mi_open_datafile now accepts 'original' file path to check if it's
an allowed symlink.
myisam/mi_static.c
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
myisam_test_invlaid_symlink defined
myisam/myisamchk.c
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
mi_open_datafile call modified
myisam/myisamdef.h
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
mi_open_datafile interface modified - 'real_path' parameter added
mysql-test/r/symlink.test
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
error codes corrected as some patch now rejected pointing inside datahome
mysql-test/r/symlink.result
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
error messages corrected in the result
mysys/my_symlink.c
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
my_is_symlink() implementsd
my_realpath() now returns the 'realpath' even if a file isn't a symlink
sql/mysql_priv.h
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
test_if_data_home_dir interface
sql/mysqld.cc
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
myisam_test_invalid_symlik set with the 'test_if_data_home_dir'
sql/sql_parse.cc
Bug#32167 another privilege bypass with DATA/INDEX DIRECTORY.
error messages corrected
test_if_data_home_dir code fixed
Send_field.org_col_name has broken value on secondary execution.
It happens when result field is created from the field which belongs to view
due to forgotten assignment of some Send_field attributes.
The fix:
set Send_field.org_col_name,org_table_name with correct value during Send_field intialization.
Length value is the length of the field,
Max_length is the length of the field value.
So Max_length can not be more than Length.
The fix: fixed calculation of the Item_empty_string item length
(Patch applied and queued on demand of Trudy/Davi.)
used causes server crash.
When the loose index scan access method is used values of aggregated functions
are precomputed by it. Aggregation of such functions shouldn't be performed
in this case and functions should be treated as normal ones.
The create_tmp_table function wasn't taking this into account and this led to
a crash if a query has MIN/MAX aggregate functions and employs temporary table
and loose index scan.
Now the JOIN::exec and the create_tmp_table functions treat MIN/MAX aggregate
functions as normal ones when the loose index scan is used.
This fix is for 5.0 only : back porting the 6.0 patch manually
The parser code in sql/sql_yacc.yy needs to be more robust to out of
memory conditions, so that when parsing a query fails due to OOM,
the thread gracefully returns an error.
Before this fix, a new/alloc returning NULL could:
- cause a crash, if dereferencing the NULL pointer,
- produce a corrupted parsed tree, containing NULL nodes,
- alter the semantic of a query, by silently dropping token values or nodes
With this fix:
- C++ constructors are *not* executed with a NULL "this" pointer
when operator new fails.
This is achieved by declaring "operator new" with a "throw ()" clause,
so that a failed new gracefully returns NULL on OOM conditions.
- calls to new/alloc are tested for a NULL result,
- The thread diagnostic area is set to an error status when OOM occurs.
This ensures that a request failing in the server properly returns an
ER_OUT_OF_RESOURCES error to the client.
- OOM conditions cause the parser to stop immediately (MYSQL_YYABORT).
This prevents causing further crashes when using a partially built parsed
tree in further rules in the parser.
No test scripts are provided, since automating OOM failures is not
instrumented in the server.
Tested under the debugger, to verify that an error in alloc_root cause the
thread to returns gracefully all the way to the client application, with
an ER_OUT_OF_RESOURCES error.
innodb-5.0-ss2475.
Bug #34286 Assertion failure in thread 2816 in file .\row\row0sel.c line 3500
Since autoinc init performs a MySQL SELECT query to determine the auto-inc
value, set prebuilt->sql_stat_start = TRUE so that it is performed like any
normal SELECT, regardless of the context in which it was invoked.
Bug #35352 If InnoDB crashes with UNDO slots full error the error persists on restart
We've added a heuristic that checks the size of the UNDO slots cache lists
(insert and upate). If either of cached lists has more than 500 entries then we
add any UNDO slots that are freed, to the common free list instead of the cache
list, this is to avoid the case where all the free slots end up in only one of
the lists on startup after a crash.
Tested with test case for 26590 and passes all mysql-test(s).
Bug #36600 SHOW STATUS takes a lot of CPU in buf_get_latched_pages_number
Fixed by removing the Innodb_buffer_pool_pages_latched variable from SHOW
STATUS output in non-UNIV_DEBUG compilation.
min() and max() functions are implemented in MySQL as macros.
This means that max(a,b) is expanded to: ((a) > (b) ? (a) : (b))
Note how 'a' is quoted two times.
Now imagine 'a' is a recursive function call that's several 10s of levels deep.
And the recursive function does max() with a function arg as well to dive into
recursion.
This means that simple function call can take most of the clock time.
Identified and fixed several such calls to max()/min() : including the IF()
sql function implementation.
Calling List<Cached_item>::delete_elements for the same list twice
caused a crash of the server in the function JOIN::cleaunup.
Ensured that delete_elements() in JOIN::cleanup would be called only once.
Range scan in descending order for c <= <col> <= c type of
ranges was ignoring the DESC flag.
However some engines like InnoDB have the primary key parts
as a suffix for every secondary key.
When such primary key suffix is used for ordering ignoring
the DESC is not valid.
But we generally would like to do this because it's faster.
Fixed by performing only reverse scan if the primary key is used.
Removed some dead code in the process.
- In QUICK_INDEX_MERGE_SELECT::read_keys_and_merge: when we got table->sort from Unique,
tell init_read_record() not to use rr_from_cache() because a) rowids are already sorted
and b) it might be that the the data is used by filesort(), which will need record rowids
(which rr_from_cache() cannot provide).
- Fully de-initialize the table->sort read in QUICK_INDEX_MERGE_SELECT::get_next(). This fixes BUG#35477.
(bk trigger: file as fix for BUG#35478).
build)
The crash was caused by freeing the internal parser stack during the parser
execution.
This occured only for complex stored procedures, after reallocating the parser
stack using my_yyoverflow(), with the following C call stack:
- MYSQLparse()
- any rule calling sp_head::restore_lex()
- lex_end()
- x_free(lex->yacc_yyss), xfree(lex->yacc_yyvs)
The root cause is the implementation of stored procedures, which breaks the
assumption from 4.1 that there is only one LEX structure per parser call.
The solution is to separate the LEX structure into:
- attributes that represent a statement (the current LEX structure),
- attributes that relate to the syntax parser itself (Yacc_state),
so that parsing multiple statements in stored programs can create multiple
LEX structures while not changing the unique Yacc_state.
Now, Yacc_state and the existing Lex_input_stream are aggregated into
Parser_state, a structure that represent the complete state of the (Lexical +
Syntax) parser.
offset for time part in UUIDs was 1/1000 of what it
should be. In other words, offset was off.
Also handle the case where we count into the future
when several UUIDs are generated in one "tick", and
then the next call is late enough for us to unwind
some but not all of those borrowed ticks.
Lastly, handle the case where we keep borrowing and
borrowing until the tick-counter overflows by also
changing into a new "numberspace" by creating a new
random suffix.
enabled)
Before this fix, the lexer and parser would treat the ';' character as a
different token (either ';' or END_OF_INPUT), based on convoluted logic,
which failed in simple cases where a stored procedure is implemented as a
single statement, and used in a multi query.
With this fix:
- the character ';' is always parsed as a ';' token in the lexer,
- parsing multi queries is implemented in the parser, in the 'query:' rules,
- the value of thd->client_capabilities, which is the capabilities
negotiated between the client and the server during bootstrap,
is immutable and not arbitrarily modified during parsing (which was the
root cause of the bug)
On certain kinds of errors (e.g., out of stack), a call to Item_func_
set_user_var::fix_fields() might fail. Since the return value of this
call was not checked inside User_var_log_event::exec_event(), continuing
execution after this will cause a crash inside Item_func_set_user_var::
update_hash().
The bug is fixed by aborting execution of the event with an error if
fix_fields() fails, since it is not possible to continue execution anyway.
