Hash index is vcol-based wrapper (MDEV-371). row_end is added to
unique index. So when row_end is updated unique hash index must be
recalculated via vcol_update_fields(). DELETE did not update virtual
fields, so DELETE HISTORY was getting wrong hash value.
The fix does update_virtual_fields() on vers_update_end() so in every
case row_end is updated virtual fields are updated as well.
Search conditions were evaluated using val_int(), which was wrong.
Fixing the code to use val_bool() instead.
Details:
- Adding a new item_base_t::IS_COND flag which marks Items used
as <search condition> in WHERE, HAVING, JOIN ON, CASE WHEN clauses.
The flag is at the parse time.
These expressions must be evaluated using val_bool() rather than val_int().
Note, the optimizer creates more Items which are used as search conditions.
Most of these items are not marked with IS_COND yet. This is OK for now,
but eventually these Items can also be fixed to have the flag.
- Adding a method Item::is_cond() which tests if the Item has the IS_COND flag.
- Implementing Item_cache_bool. It evaluates the cached expression using
val_bool() rather than val_int().
Overriding Type_handler_bool::Item_get_cache() to create Item_cache_bool.
- Implementing Item::save_bool_in_field(). It uses val_bool() rather than
val_int() to evaluate the expression.
- Implementing Type_handler_bool::Item_save_in_field()
using Item::save_bool_in_field().
- Fixing all Item_bool_func descendants to implement a virtual val_bool()
rather than a virtual val_int().
- To find places where val_int() should be fixed to val_bool(), a few
DBUG_ASSERT(!is_cond()) where added into val_int() implementations
of selected (most frequent) classes:
Item_field
Item_str_func
Item_datefunc
Item_timefunc
Item_datetimefunc
Item_cache_bool
Item_bool_func
Item_func_hybrid_field_type
Item_basic_constant descendants
- Fixing all places where DBUG_ASSERT() happened during an "mtr" run
to use val_bool() instead of val_int().
Problem was that wsrep_schema tables were not marked as
category information. Fix allows access to wsrep_schema
tables even when node is detached.
This is 10.4-10.9 version of fix.
Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>
Discovered this while working on MDEV-34720: test_if_cheaper_ordering()
uses rec_per_key, while the original estimate for the access method
is produced in best_access_path() by using actual_rec_per_key().
Make test_if_cheaper_ordering() also use actual_rec_per_key().
Also make several getter function "const" to make this compile.
Also adjusted the testcase to handle this (the change backported from
11.0)
Commit a8a75ba2d causes the MariaDB server to crash, usually with signal
11, at random code locations due to invalid pointer values during any
table operation. This issue occurs when the server is built with -O3 and
other customized compiler flags.
For example, the command `use db1;` causes server to crash in the
`check_table_access` function at line sql_parse.cc:7080 because
`tables->correspondent_table` is an invalid pointer value of 0x1.
The crashes are due to undefined behavior from using uninitialized
variables. The problematic commit a8a75ba2d introduces code that
allocates memory and sets it to 0 using thd->calloc before initializing
it with a placement new operation.
This process depends on setting memory to 0 to initialize member
variables not explicitly set in the constructor. However, the compiler
can optimize out the memset/bfill, leading to uninitialized values and
unpredictable issues.
Once a constructor function initializes an object, any uninitialized
variables within that object are subject to undefined behavior. The
state of memory before the constructor runs, whether it involves
memset or was used for other purposes, is irrelevant after the
placement new operation.
This behavior can be demonstrated with this
[test](https://gcc.godbolt.org/z/5n87z1raG) I wrote to examine the
assembly code. The code in MariaDB can be abstracted to the following,
though it has many layers wrapped around it and more complex logic,
causing slight differences in optimization in the MariaDB build.
To summarize, on x86, the memset in the following code is optimized out
with both -O2 and -O3 in GCC 13, and is only preserved in the much older
GCC 4.9.
struct S {
int i; // uninitialized in consturctor
S() {};
};
int bar() {
void *buf = malloc(sizeof(S));
memset(buf, 0, sizeof(S)); // optimized out
S* s = new(buf) S;
return s->i;
}
With GCC13 -O3:
bar():
sub rsp, 8
mov edi, 4
call malloc
mov eax, DWORD PTR [rax]
add rsp, 8
ret
With GCC4.9 -O3
bar():
sub rsp, 8
mov edi, 4
call malloc
mov DWORD PTR [rax], 0
xor eax, eax
add rsp, 8
ret
Now we ensure the constructor initializes variables correctly by running
the reset() function in the constructor to perform the memset/bfill(0)
operation. After applying the fix, the crash is gone.
All new code of the whole pull request, including one or several files
that are either new files or modified ones, are contributed under the
BSD-new license. I am contributing on behalf of my employer Amazon Web
Services.
Improve performance of queries like
SELECT * FROM t1 WHERE field = NAME_CONST('a', 4);
by, in this example, replacing the WHERE clause with field = 4
in the case of ref access.
The rewrite is done during fix_fields and we disambiguate this
case from other cases of NAME_CONST by inspecting where we are
in parsing. We rely on THD::where to accomplish this. To
improve performance there, we change the type of THD::where to
be an enumeration, so we can avoid string comparisons during
Item_name_const::fix_fields. Consequently, this patch also
changes all usages of THD::where to conform likewise.
The patch for MDEV-31340 fixed the following bugs:
MDEV-33084 LASTVAL(t1) and LASTVAL(T1) do not work well with lower-case-table-names=0
MDEV-33085 Tables T1 and t1 do not work well with ENGINE=CSV and lower-case-table-names=0
MDEV-33086 SHOW OPEN TABLES IN DB1 -- is case insensitive with lower-case-table-names=0
MDEV-33088 Cannot create triggers in the database `MYSQL`
MDEV-33103 LOCK TABLE t1 AS t2 -- alias is not case sensitive with lower-case-table-names=0
MDEV-33108 TABLE_STATISTICS and INDEX_STATISTICS are case insensitive with lower-case-table-names=0
MDEV-33109 DROP DATABASE MYSQL -- does not drop SP with lower-case-table-names=0
MDEV-33110 HANDLER commands are case insensitive with lower-case-table-names=0
MDEV-33119 User is case insensitive in INFORMATION_SCHEMA.VIEWS
MDEV-33120 System log table names are case insensitive with lower-cast-table-names=0
Backporting the fixes from 11.5 to 10.5
on disable_indexes(HA_KEY_SWITCH_NONUNIQ_SAVE) the engine does
not know that the long unique is logically unique, because on the
engine level it is not. And the engine disables it,
Change the disable_indexes/enable_indexes API. Instead of the enum
mode, send a key_map of indexes that should be enabled. This way the
server will decide what is unique, not the engine.
Ideally our methods and functions should do one thing, do that well,
and do only that. add_table_to_list does far more than adding a
table to a list, so this commit factors the TABLE_LIST creation out
to a new TABLE_LIST constructor. It then uses placement new()
to create it in the correct memory area (result of thd->calloc).
Benefits of this approach:
1. add_table_to_list now returns as early as possible on an error
2. fewer side-effects incurred on creating the TABLE_LIST object
3. TABLE_LIST won't be calloc'd if copy_to_db fails
4. local declarations moved closer to their respective first uses
5. improved code readability and logical flow
Also factored a couple of other functions to keep the happy path
more to the left, which makes them easier to follow at a glance.
According to the standard, the autoincrement column (i.e. *identity
column*) should be advanced each insert implicitly made by
UPDATE/DELETE ... FOR PORTION.
This is very unconvenient use in several notable cases. Concider a
WITHOUT OVERLAPS key with an autoinc column:
id int auto_increment, unique(id, p without overlaps)
An update or delete with FOR PORTION creates a sense that id will remain
unchanged in such case.
The standard's IDENTITY reminds MariaDB's AUTO_INCREMENT, however
the generation rules differ in many ways. For example, there's also a
notion autoincrement index, which is bound to the autoincrement field.
We will define our own generation rule for the PORTION OF operations
involving AUTO_INCREMENT:
* If an autoincrement index contains WITHOUT OVERLAPS specification, then
a new value should not be generated, otherwise it should.
Apart from WITHOUT OVERLAPS there is also another notable case, referred
by the reporter - a unique key that has an autoincrement column and a field
from the period specification:
id int auto_increment, unique(id, s), period for p(s, e)
for this case, no exception is made, and the autoincrementing rules will be
proceeded accordung to the standard (i.e. the value will be advanced on
implicit inserts).
use the original, not the truncated, field in the long unique prefix,
that is, in the hash(left(field, length)) expression.
because MyISAM CHECK/REPAIR in compute_vcols() moves table->field
but not prefix fields from keyparts.
Also, implement Field_string::cmp_prefix() for prefix comparison
of CHAR columns to work.
Most things where wrong in the test suite.
The one thing that was a bug was that table_map_id was in some places
defined as ulong and in other places as ulonglong. On Linux 64 bit this
is not a problem as ulong == ulonglong, but on windows this caused failures.
Fixed by ensuring that all instances of table_map_id are ulonglong.
This is the follow-up patch that removes explicit use of thd->stmt_arena
for memory allocation and replaces it with call of the method
THD::active_stmt_arena_to_use()
Additionally, this patch adds extra DBUG_ASSERT to check that right
query arena is in use.
Attempt to resolve FOR SYSTEM_TIME expression as field for derived
table is done before derived table is fully prepared, so we fail on
assertion that table_list->table is missing.
Actually Vers_history_point::resolve_unit() is done under the call of
mysql_derived_prepare() itself (sql_derived.cc:824) and the table is
assigned later at 867.
The fix disables unit resolution for field type in FOR SYSTEM_TIME
expression as it does a little sense in any case: making historical
queries based on variable field values produces the result of multiple
time points.
fix_fields_if_needed() in resolve_units() was introduced by 46be31982a
remove the hack where NO_DEFAULT_VALUE_FLAG was temporarily removed
from a field to initialize DEFAULT() functions in CHECK constraints
while disabling self-reference field checks.
Instead, initialize DEFAULT() functions in CHECK explicitly,
don't call check_field_expression_processor() for CHECK at all.
32 bit MariaDB crashed in innodb.innodb-16k and a few other tests.
Fixed by using correct sizeof() calls.
Histograms where not read if first read was without histograms.
This commits enables reloading of engine-independent statistics
without flushing the table from table definition cache.
This is achieved by allowing multiple version of the
TABLE_STATISTICS_CB object and having independent pointers to it in
TABLE and TABLE_SHARE. The TABLE_STATISTICS_CB object have reference
pointers and are freed when no one is pointing to it anymore.
TABLE's TABLE_STATISTICS_CB pointer is updated to use the
TABLE_SHARE's pointer when read_statistics_for_tables() is called at
the beginning of a query.
Main changes:
- read_statistics_for_table() will allocate an new TABLE_STATISTICS_CB
object.
- All get_stat_values() functions has a new parameter that tells
where collected data should be stored. get_stat_values() are not
using the table_field object anymore to store data.
- All get_stat_values() functions returns 1 if they found any
data in the statistics tables.
Other things:
- Fixed INSERT DELAYED to not read statistics tables.
- Removed Statistics_state from TABLE_STATISTICS_CB as this is not
needed anymore as wer are not changing TABLE_SHARE->stats_cb while
calculating or loading statistics.
- Store values used with store_from_statistical_minmax_field() in
TABLE_STATISTICS_CB::mem_root. This allowed me to remove the function
delete_stat_values_for_table_share().
- Field_blob::store_from_statistical_minmax_field() is implemented
but is not normally used as we do not yet support EIS statistics
for blobs. For example Field_blob::update_min() and
Field_blob::update_max() are not implemented.
Note that the function can be called if there is an concurrent
"ALTER TABLE MODIFY field BLOB" running because of a bug in
ALTER TABLE where it deletes entries from column_stats
before it has an exclusive lock on the table.
- Use result of field->val_str(&val) as a pointer to the result
instead of val (safetly fix).
- Allocate memory for collected statistics in THD::mem_root, not in
in TABLE::mem_root. This could cause the TABLE object to grow if a
ANALYZE TABLE was run many times on the same table.
This was done in allocate_statistics_for_table(),
create_min_max_statistical_fields_for_table() and
create_min_max_statistical_fields_for_table_share().
- Store in TABLE_STATISTICS_CB::stats_available which statistics was
found in the statistics tables.
- Removed index_table from class Index_prefix_calc as it was not used.
- Added TABLE_SHARE::LOCK_statistics to ensure we don't load EITS
in parallel. First thread will load it, others will reuse the
loaded data.
- Eliminate read_histograms_for_table(). The loading happens within
read_statistics_for_tables() if histograms are needed.
One downside is that if we have read statistics without histograms
before and someone requires histograms, we have to read all statistics
again (once) from the statistics tables.
A smaller downside is the need to call alloc_root() for each
individual histogram. Before we could allocate all the space for
histograms with a single alloc_root.
- Fixed bug in MyISAM and Aria where they did not properly notice
that table had changed after analyze table. This was not a problem
before this patch as then the MyISAM and Aria tables where flushed
as part of ANALYZE table which did hide this issue.
- Fixed a bug in ANALYZE table where table->records could be seen as 0
in collect_statistics_for_table(). The effect of this unlikely bug
was that a full table scan could be done even if
analyze_sample_percentage was not set to 1.
- Changed multiple mallocs in a row to use multi_alloc_root().
- Added a mutex protection in update_statistics_for_table() to ensure
that several tables are not updating the statistics at the same time.
Some of the changes in sql_statistics.cc are based on a patch from
Oleg Smirnov <olernov@gmail.com>
Co-authored-by: Oleg Smirnov <olernov@gmail.com>
Co-authored-by: Vicentiu Ciorbaru <cvicentiu@gmail.com>
Reviewer: Sergei Petrunia <sergey@mariadb.com>
ANALYZE FORMAT=JSON output now includes table.r_engine_stats which
has the engine statistics. Only non-zero members are printed.
Internally: EXPLAIN data structures Explain_table_acccess and
Explain_update now have handler* handler_for_stats pointer.
It is used to read statistics from handler_for_stats->handler_stats.
The following applies only to 10.9+, backport doesn't use it:
Explain data structures exist after the tables are closed. We avoid
walking invalid pointers using this:
- SQL layer calls Explain_query::notify_tables_are_closed() before
closing tables.
- After that call, printing of JSON output is disabled. Non-JSON output
can be printed but we don't access handler_for_stats when doing that.
The new statistics is enabled by adding the "engine", "innodb" or "full"
option to --log-slow-verbosity
Example output:
# Pages_accessed: 184 Pages_read: 95 Pages_updated: 0 Old_rows_read: 1
# Pages_read_time: 17.0204 Engine_time: 248.1297
Page_read_time is time doing physical reads inside a storage engine.
(Writes cannot be tracked as these are usually done in the background).
Engine_time is the time spent inside the storage engine for the full
duration of the read/write/update calls. It uses the same code as
'analyze statement' for calculating the time spent.
The engine statistics is done with a generic interface that should be
easy for any engine to use. It can also easily be extended to provide
even more statistics.
Currently only InnoDB has counters for Pages_% and Undo_% status.
Engine_time works for all engines.
Implementation details:
class ha_handler_stats holds all engine stats. This class is included
in handler and THD classes.
While a query is running, all statistics is updated in the handler. In
close_thread_tables() the statistics is added to the THD.
handler::handler_stats is a pointer to where statistics should be
collected. This is set to point to handler::active_handler_stats if
stats are requested. If not, it is set to 0.
handler_stats has also an element, 'active' that is 1 if stats are
requested. This is to allow engines to avoid doing any 'if's while
updating the statistics.
Cloned or partition tables have the pointer set to the base table if
status are requested.
There is a small performance impact when using --log-slow-verbosity=engine:
- All engine calls in 'select' will be timed.
- IO calls for InnoDB reads will be timed.
- Incrementation of counters are done on local variables and accesses
are inline, so these should have very little impact.
- Statistics has to be reset for each statement for the THD and each
used handler. This is only 40 bytes, which should be neglectable.
- For partition tables we have to loop over all partitions to update
the handler_status as part of table_init(). Can be optimized in the
future to only do this is log-slow-verbosity changes. For this to work
we have to update handler_status for all opened partitions and
also for all partitions opened in the future.
Other things:
- Added options 'engine' and 'full' to log-slow-verbosity.
- Some of the new files in the test suite comes from Percona server, which
has similar status information.
- buf_page_optimistic_get(): Do not increment any counter, since we are
only validating a pointer, not performing any buf_pool.page_hash lookup.
- Added THD argument to save_explain_data_intern().
- Switched arguments for save_explain_.*_data() to have
always THD first (generates better code as other functions also have THD
first).
mysql.proc. The table is probably corrupted"
Analysis: When mysql_upgrade runs statements for upgrade, characterset is
converted to utf8mb4 because server starts with old_mode that interprets
utf8 to utf8mb4, but mysql.proc table has "utf8mb3" as hardcoded, so
it crashes with corrupted table.
Fix: Changed Table_check_intact::check() definition to allow both
utf8mb3 and utf8mb4 by checking prefix and changing the upgrade scripts
to explicitly use utf8mb3
MDEV-29253 Detect incompatible MySQL partition scheme and either convert
them or report to user and in error log.
This task is about converting in place MySQL 5.6 and 5.7 partition tables
to MariaDB as part of mariadb-upgrade.
- Update TABLE_SHARE::init_from_binary_frm_image() to be able to read
MySQL frm files with partitions.
- Create .par file, if it do not exists, on open of partitioned table.
Executing mariadb-upgrade will create all the missing .par files.
The MySQL .frm file will be changed to MariaDB format after next
ALTER TABLE.
Other changes:
- If we are using stored mysql_version to distingush between MySQL and
MariaDB .frm file information, do not upgrade mysql_version in the
.frm file as part of CHECK TABLE .. FOR UPGRADE as this would cause
problems next time we parse the .frm file.
- Moved view checks after privilege tables are fixed. This is to avoid
warnings about wrongly defined mysql.proc when checking views.
- Don't use stat tables before they have been fixed.
- Don't run mysql_fix_view() if 'FOR MYSQL' is used if the view is
already a MariaDB view.
- Added 'FOR UPGRADE' as an option for 'REPAIR VIEW' to be able to
detect if the REPAIR command comes from mariadb_upgrade. In this
case we get a warning, instead of an error, if a definer of a view
does not exists.
Set mysql.wsrep_cluster and mysql.wsrep_cluster_members as
TABLE_CATEGORY_INFORMATION as mysql.wsrep_streaming_log
so that they can be queried even if node is not primary
component.
Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>
This bug could cause a crash of the server when processing a query with
ROWNUM() if it used in its FROM list a reference to a mergeable view
defined as SELECT over more than one table that contained ORDER BY clause.
When a mergeable view with ORDER BY clause and without LIMIT clause is used
in the FROM list of a query that does not have ORDER BY clause the ORDER BY
clause of the view is moved to the query. The code that performed this
transformation forgot to delete the moved ORDER BY list from the view.
If a query contains ROWNUM() and uses a mergeable multi-table view with
ORDER BY then according to the current code of TABLE_LIST::init_derived()
the view has to be forcibly materialized. As the query and the view shared
the same items in its ORDER BY lists they could not be properly resolved
either in the query or in the view. This led to a crash of the server.
This patch has returned back the original signature of LEX::can_not_use_merged()
to comply with 10.4 code of the condition that checks whether a megeable
view has to be forcibly materialized.
Approved by Oleksandr Byelkin <sanja@mariadb.com>
