Remove the HA_NO_BLOBS restriction from the HEAP engine, allowing
the optimizer to keep temporary tables with BLOB/TEXT columns in
memory when they fit within max_heap_table_size / tmp_memory_table_size
limits. Additionally, advertise HA_CAN_GEOMETRY so explicit
CREATE TABLE ... ENGINE=MEMORY with GEOMETRY columns works.
Unlike other HEAP blob implementations (e.g. Percona), this patch
provides full HASH index support on blob columns, enabling efficient
lookups, GROUP BY, and DISTINCT operations directly in HEAP without
falling back to disk.
Architecture
------------
BLOB data is stored using continuation records -- additional fixed-size
records allocated from the same HP_BLOCK that holds regular rows. This
reuses existing allocation, free list, and size accounting with minimal
structural change, and avoids per-blob my_malloc() calls.
The existing single-byte visibility flag is extended into a flags byte
with bits for HP_ROW_HAS_CONT, HP_ROW_IS_CONT, HP_ROW_CONT_ZEROCOPY,
HP_ROW_SINGLE_REC, and HP_ROW_MULTIPLE_REC. Continuation records are
grouped into variable-length runs -- contiguous sequences within a leaf
block. Only the first record of each run carries a 10-byte header
(next_cont pointer + run_rec_count); inner records are pure payload.
Three storage formats, detected by flag bits via inline predicates:
Case A (HP_ROW_SINGLE_REC): single record, no header, data at
offset 0. Zero-copy read.
Case B (HP_ROW_CONT_ZEROCOPY): single run, multiple records.
Header in rec 0, data contiguous in rec 1..N-1. Zero-copy read
via chain + recbuffer.
Case C (HP_ROW_MULTIPLE_REC): one or more runs linked via
next_cont. Reassembly into blob_buff required.
Run allocation uses a two-phase strategy: (1) peek-then-unlink walk
of the free list detecting contiguous groups, (2) tail allocation
from HP_BLOCK for remaining data. A Step 3 scavenge fallback
walks the entire free list when tail allocation fails.
HP_SHARE::total_records tracks all physical records (primary +
continuation), while HP_SHARE::records remains the logical count
used by hash bucket mapping.
Reassembly buffer (HP_INFO::blob_buff) follows the same pattern as
InnoDB's blob_heap -- allocated once, grown via my_realloc, freed
on heap_reset()/close. Zero-copy cases (A/B) return pointers
directly into HP_BLOCK with no copy.
Full HASH index key handling for BLOB columns: hp_rec_hashnr(),
hp_rec_key_cmp(), hp_key_cmp(), hp_make_key(), hp_hashnr() are
extended for HA_BLOB_PART segments. Hash pre-check optimization
skips expensive blob materialization when hashes differ. PAD SPACE
collation semantics are preserved for blob key comparisons.
Field_blob_key (Monty) produces HEAP-native key format (4-byte length
+ 8-byte data pointer) directly, eliminating key buffer translation
between the SQL layer and HEAP engine.
SQL layer changes
-----------------
pick_engine() (new, extracted from choose_engine()): replaced the
blob_fields check with a reclength > HA_MAX_REC_LENGTH guard.
choose_engine() calls pick_engine() with the real reclength.
pick_engine() is also called early in finalize() with reclength=0
to predict whether the engine will be HEAP, enabling blob-aware
GROUP BY key setup that avoids unnecessary m_using_unique_constraint.
finalize(): HEAP+blob uses fixed-width rows; GROUP BY key setup sets
key_part_flag from field, uses item max_length for blob key sizing.
store_length initialized for all GROUP BY key parts. key_type uses
field->binary() to determine FIELDFLAG_BINARY vs text collation.
DISTINCT key setup skips null-bits helper for HEAP.
remove_duplicates(): blob check moved before HEAP check to fall
through to remove_dup_with_compare().
Aggregator_distinct::add(): overflow-to-disk conversion via
create_internal_tmp_table_from_heap() for non-dup write errors.
Expression cache disabled for HEAP+blob (key format incompatibility).
FULLTEXT early detection in mysql_derived_prepare(): forces disk
engine via TMP_TABLE_FORCE_MYISAM when outer query uses MATCH.
Deferred blob chain free (MDEV-39732): heap_delete() saves chain
pointers to pending_blob_chains, flushed on next mutation or
heap_reset()/close. Prevents dangling zero-copy pointers during
binlog_log_row().
REPLACE safety (MDEV-39825): HP_SHARE::write_can_replace flag
forces copy mode in hp_read_blobs(), preventing blob data corruption
from freed-then-reused continuation records during REPLACE.
Geometry GROUP_CONCAT fix (MDEV-39761): downgrade Field_geom to
Field_blob for GROUP_CONCAT temp tables in both expression creation
paths. Type_handler_geometry::type_handler_for_tmp_table() added.
Geometry GROUP BY key fix (MDEV-39871): detect when new_key_field()
produced non-blob Field_varstring for a blob column, replace with
Field_blob_key.
Performance
-----------
Non-blob tables: zero regression. Every blob-specific code path is
guarded by if (share->blob_count). No new allocations, no format
changes, no hash function changes for non-blob keys.
Blob tables: eliminates file creation/deletion overhead and page cache
management. For single-run blobs (common case), the read path is
entirely zero-copy.
Limitations
-----------
1. No BTREE indexes on blob columns (HASH only)
2. No partial-key prefix indexing for blobs
3. 2x memory for Case C reads only (A/B are 1x)
4. No blob compression
5. 65,535 records per run (uint16 cap, auto-split)
6. max_heap_table_size applies to continuation records
7. Expression cache disabled for HEAP+blob
8. FULLTEXT forces disk engine
Linked bugs fixed:
- MDEV-39703: mroonga fulltext test ordering
- MDEV-39723: ER_DUP_ENTRY on GROUP BY with blob column
- MDEV-39724: crash in hp_is_single_rec with GROUP BY
- MDEV-39732: slave crash in hp_free_run_chain on blob replication
- MDEV-39761: Field_geom::store() assertion in GROUP_CONCAT
- MDEV-39782: RBR ER_KEY_NOT_FOUND on HEAP blob UPDATE
- MDEV-39825: blob data corruption on REPLACE into HEAP table
- MDEV-39871: crash in my_hash_sort_bin on GROUP BY with geometry
Reviewed by: Michael Widenius <monty@mariadb.org>
Monty reviewed the entire patch. Areas where he suggested changes
or contributed code:
- Field_blob_key class (HEAP-native blob key format, 4-byte length +
data pointer)
- Duplicate key fix on HEAP-to-Aria conversion
- hp_blob_key_length() uint32 fix
- hp_rec_hashnr_stored removal
- type_handler_for_tmp_table() param cleanup
- Type_handler_geometry::type_handler_for_tmp_table() virtual
- blob pointer bzero()
- find_unique_row() double-materialization fix
- Tail reclaim review
- Batch tail allocation review
- hp_update.c cleanup
- Field_blob_compressed temp table fix
- row_pack_length() dedup
- pack_length_no_ptr() removal
- Race condition fix in HEAP
- MDEV-39703 mroonga test fix
- MDEV-39825 write_can_replace optimization
- is_text_key_segment removal (field->binary() simplification)
- Documentation (Docs/internal-temporary-tables.txt)
Contribution by: Alexander Barkov <bar@mariadb.com>
Type_handler::make_and_init_table_field_ex() -- refactored temp table
field creation from inline code in sql_select.cc into type handler
virtual methods (sql_type.cc, sql_type_geom.cc), enabling clean
per-type-handler field creation for HEAP blob promotion.
A prepared statement runs preparation again on every execution, and
preparation merges any VIEWs named in the statement. A VIEW that
appears only inside a subquery is meant to be skipped by the merge for
insert pass during mysql_handle_derived, since it is not the target of
the DELETE.
The first execution prepares with the subquery still nested, so the
view is correctly skipped, and the DELETE succeeds. Optimization then
converts the IN subquery into a semi-join and adds its tables into the
first SELECT_LEX's table list (reassigning them to that SELECT_LEX).
This conversion is meant to persist for later executions.
Then the PS's second execution prepares against the already flattened
VIEW. mysql_handle_derived now finds the subquery's VIEW among the
first SELECT_LEX's tables and tries to merge it. mysql_handle_derived
already had some existing skip logic to recognize a VIEW, but the
semijoin conversion has already merged the VIEW into the first
SELECT_LEX, so the skip doesn't correctly apply. So a nested VIEW is
merged for insert with no table, causing a crash on the NULL table
pointer.
Solution is to skip a table during mysql_handle_derived when it is the
inner side of a semijoin, recognized by its embedding semijoin nest.
Outdated table_list->derived_result after first trigger run. It was
allocated on statment mem_root and freed when trigger procedure
finishes.
Next INSERT trigger on new view in mysql_derived_prepare() jumps to
exit on error before it had a chance to allocate new derived_result.
It jumps here:
797 if ((res= sl->handle_derived(lex, DT_PREPARE)))
798 goto exit;
and the allocation is later:
837 if ((!derived->is_with_table_recursive_reference() ||
838 !derived->derived_result) &&
839 !(derived->derived_result= new (thd->mem_root) select_unit(thd)))
840 DBUG_RETURN(TRUE); // out of memory
on exit it does destruction of already freed object from previous statement:
901 exit:
...
921 if (res)
922 {
923 if (!derived->is_with_table_recursive_reference())
924 {
...
927 delete derived->derived_result;
The most obvious fix is to cleanup derived_result when the TABLE_LIST
is reinited for new use. Also as a precaution the patch does the
cleanup after delete operator.
Rename cloning methods of class Item and its descendants
in the following way:
(from) (to)
do_build_clone -> deep_copy
build_clone -> deep_copy_with_checks
do_get_copy -> shallow_copy
get_copy -> shallow_copy_with_checks
to better reflect their functionality.
Also make Item::deep_copy() and shallow_copy() protected.
Outside users should call deep_copy_with_checks()
and shallow_copy_with_checks().
Variant 2: "Don't call optimize_stage2 too early"
The affected query had Split-Materialized derived table in another
derived table:
select * -- select#1
from (
select * -- select#2
from t1,
(select * from t2 ... group by t2.group_col) DT -- select#3
where t1.col=t2.group_col
) TBL;
The optimization went as follows:
JOIN::optimize() calls select_lex->handle_derived(DT_OPTIMIZE) which
calls JOIN::optimize() for all (direct and indirect) children SELECTs.
select#1->handle_derived() calls JOIN::optimize() for select#2 and #3;
select#2->handle_derived() calls JOIN::optimize() for select#3 the second
time. That call would call JOIN::optimize_stage2(), assuming the query
plan choice has been finalized for select#3.
But after that, JOIN::optimize() for select#2 would continue and consider
Split-Materialized for select#3. This would attempt to pick another join
order and cause a crash.
The fix is to have JOIN::optimize() not call optimize_stage2() ever.
Finalizing the query plan choice is now done by a separate call:
select_lex->handle_derived(thd->lex, DT_OPTIMIZE_STAGE2)
which invokes JOIN::optimize_stage2() and saves the query plan.
This patch adds support for SYS_REFCURSOR (a weakly typed cursor)
for both sql_mode=ORACLE and sql_mode=DEFAULT.
Works as a regular stored routine variable, parameter and return value:
- can be passed as an IN parameter to stored functions and procedures
- can be passed as an INOUT and OUT parameter to stored procedures
- can be returned from a stored function
Note, strongly typed REF CURSOR will be added separately.
Note, to maintain dependencies easier, some parts of sql_class.h
and item.h were moved to new header files:
- select_results.h:
class select_result_sink
class select_result
class select_result_interceptor
- sp_cursor.h:
class sp_cursor_statistics
class sp_cursor
- sp_rcontext_handler.h
class Sp_rcontext_handler and its descendants
The implementation consists of the following parts:
- A new class sp_cursor_array deriving from Dynamic_array
- A new class Statement_rcontext which contains data shared
between sub-statements of a compound statement.
It has a member m_statement_cursors of the sp_cursor_array data type,
as well as open cursor counter. THD inherits from Statement_rcontext.
- A new data type handler Type_handler_sys_refcursor in plugins/type_cursor/
It is designed to store uint16 references -
positions of the cursor in THD::m_statement_cursors.
- Type_handler_sys_refcursor suppresses some derived numeric features.
When a SYS_REFCURSOR variable is used as an integer an error is raised.
- A new abstract class sp_instr_fetch_cursor. It's needed to share
the common code between "OPEN cur" (for static cursors) and
"OPER cur FOR stmt" (for SYS_REFCURSORs).
- New sp_instr classes:
* sp_instr_copen_by_ref - OPEN sys_ref_curor FOR stmt;
* sp_instr_cfetch_by_ref - FETCH sys_ref_cursor INTO targets;
* sp_instr_cclose_by_ref - CLOSE sys_ref_cursor;
* sp_instr_destruct_variable - to destruct SYS_REFCURSOR variables when
the execution goes out of the BEGIN..END block
where SYS_REFCURSOR variables are declared.
- New methods in LEX:
* sp_open_cursor_for_stmt - handles "OPEN sys_ref_cursor FOR stmt".
* sp_add_instr_fetch_cursor - "FETCH cur INTO targets" for both
static cursors and SYS_REFCURSORs.
* sp_close - handles "CLOSE cur" both for static cursors and SYS_REFCURSORs.
- Changes in cursor functions to handle both static cursors and SYS_REFCURSORs:
* Item_func_cursor_isopen
* Item_func_cursor_found
* Item_func_cursor_notfound
* Item_func_cursor_rowcount
- A new system variable @@max_open_cursors - to limit the number
of cursors (static and SYS_REFCURSORs) opened at the same time.
Its allowed range is [0-65536], with 50 by default.
- A new virtual method Type_handler::can_return_bool() telling
if calling item->val_bool() is allowed for Items of this data type,
or if otherwise the "Illegal parameter for operation" error should be raised
at fix_fields() time.
- New methods in Sp_rcontext_handler:
* get_cursor()
* get_cursor_by_ref()
- A new class Sp_rcontext_handler_statement to handle top level statement
wide cursors which are shared by all substatements.
- A new virtual method expr_event_handler() in classes Item and Field.
It's needed to close (and make available for a new OPEN)
unused THD::m_statement_cursors elements which do not have any references
any more. It can happen in various moments in time, e.g.
* after evaluation parameters of an SQL routine
* after assigning a cursor expression into a SYS_REFCURSOR variable
* when leaving a BEGIN..END block with SYS_REFCURSOR variables
* after setting OUT/INOUT routine actual parameters from formal
parameters.
This patch includes a few changes to make the code easier to maintain:
- Renamed SQL_I_List::link_in_list to SQL_I_List::insert. link_in_list was
ambiguous as it could refer to a link or it could refer to a node
- Remove field_name local variable in multi_update::initialize_tables because
it is not used when creating the temporary tables
- multi_update changes:
- Move temp table callocs to init, a more natural location for them, and moved
tables_to_update to const member variable so we don't recompute it.
- Filter out jtbm tables and tables not in the update map, pushing those that
will be updated into an update_targets container. This simplifies checks and
loops in initialize_tables.
create templates
thd->alloc<X>(n) to use instead of (X*)thd->alloc(sizeof(X)*n)
and the same for thd->calloc(). By the default the type is char,
so old usage of thd->alloc(size) works too.
Fix for MDEV-31466 - add optional derived table column names.
Column names within a SELECT_LEX structure can be left in a non-reparsable
state (as printed out from *::print) after JOIN::prepare. This caused
an incorrect view definition to be written into the .FRM file.
Fixed by resetting item list names in SELECT_LEX structures representing
derived tables before writing out the view definition.
Reviewed by Igor Babaev (igor@mariadb.com)
Extend derived table syntax to support column name assignment.
(subquery expression) [as|=] ident [comma separated column name list].
Prior to this patch, the optional comma separated column name list is
not supported.
Processing within the unit of the subquery expression will use
original column names, outside the unit will use the new names.
For example, in the query
select a1, a2 from
(select c1, c2, c3 from t1 where c2 > 0) as dt (a1, a2, a3)
where a2 > 10;
we see the second column of the derived table dt being used both within,
(where c2 > 0), and outside, (where a2 > 10), the specification.
Both conditions apply to t1.c2.
When multiple unit preparations are required, such as when being used within
a prepared statement or procedure, original column names are needed for
correct resolution. Original names are reset within mysql_derived_reinit().
Item_holder items, used for result tables in both TVC and union preparations
are renamed before use within st_select_lex_unit::prepare().
During wildcard expansion, if column names are present, items names are
set directly after creation.
Reviewed by Igor Babaev (igor@mariadb.com)
Statements affected by this bug need all the following to be true
1) a derived table table or view whose specification contains a set
operation at the top level.
2) a grouping operator (group by/having) operating on a column alias
other than in the first select of the union/intersect
3) an outer condition that will be pushed into all selects in this
union/intersect, either into the where or having clause
When pushing a condition into all selects of a unit with more than one
select, pushdown_cond_for_derived() renames items so we can re-use the
condition being pushed.
These names need to be saved and reset for correct name resolution on
second execution of prepared statements.
Reviewed by Igor Babaev (igor@mariadb.com)
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.
For queries with derived tables populated having some side-effect, we
will fill such a derived table more than once, but without clearing
its rows. Consequently it will have duplicate rows.
An example query exhibiting the problem is
SELECT STRAIGHT_JOIN c1 FROM t1 JOIN (SELECT @a := 0) x;
Since mysql_derived_fill will, for UNCACHEABLE_DEPENDENT tables, drop
all rows and repopulate, we relax the condition at line 1204: rather
than assume all uncacheable values prevent early return, we now
allow an early return for uncacheable values other than
UNCACHEABLE_DEPENDENT. In general, we only populate derived tables
once unless they're dependent tables.
This bug led to wrong result sets returned by the second execution of
prepared statements from selects using mergeable derived tables pushed
into external engine. Such derived tables are always materialized. The
decision that they have to be materialized is taken late in the function
mysql_derived_optimized(). For regular derived tables this decision is
usually taken at the prepare phase. However in some cases for some derived
tables this decision is made in mysql_derived_optimized() too. It can be
seen in the code of mysql_derived_fill() that for such a derived table it's
critical to change its translation table to tune it to the fields of the
temporary table used for materialization of the derived table and this
must be done after each refill of the derived table. The same actions are
needed for derived tables pushed into external engines.
Approved by Oleksandr Byelkin <sanja@mariadb.com>
Consider this query
SELECT t1.* FROM t1, (SELECT t2.b FROM t2 WHERE NOT EXISTS
(SELECT 1 FROM t3) GROUP BY b) sq where sq.b = t1.a;
If SELECT 1 FROM t3 is expensive, for example t3 has >
thd->variables.expensive_subquery_limit, first evaluation is deferred to
mysql_derived_fill(). There it is noted that, in the above case
NOT EXISTS (SELECT 1 FROM t3) is constant and false.
This causes the join variable zero_result_cause to be set to
"Impossible WHERE noticed after reading const tables" and the handler
for this join is never "opened" via handler::ha_open.
When mysql_derived_fill() is called for the next group of results, this
unopened handler is not taken into account.
reviewed by Igor Babaev (igor@mariadb.com)
