Commit graph

175548 commits

Author SHA1 Message Date
Sergei Golubchik
8938031bc7 InnoDB: don't stop purge threads if there's work to do
in slow shutdown mode don't stop purge threads until they've
purged everything there is
2016-12-15 09:35:25 +01:00
Sergei Golubchik
8d770859c9 InnoDB purge thread and other bg threads
in slow shutdown mode stop all bg threads that might generate
new undo records to purge before stopping purge threads.
2016-12-15 09:34:37 +01:00
Sergei Golubchik
eabb0aef12 sporadic crashes of innodb.innodb_prefix_index_restart_server
in slow shutdown mode purge threads really must exit only when there is
nothing to purge. Restore the trx_commit_disallowed check and
don't stop purge threads until all connection thread transactions
are gone.
2016-12-15 09:33:59 +01:00
Daniel Black
a8e0c6fd9b Travis: add refs for future capability - when travis catches up 2016-12-15 10:36:11 +04:00
Daniel Black
e14bdcb81c travis: gcc-5 and gcc-6 2016-12-15 10:36:11 +04:00
Elena Stepanova
9213ac8fd8 Follow-up for a411d7f4f6 - change in formatting of SHOW CREATE TABLE 2016-12-15 02:35:31 +02:00
Elena Stepanova
a1833ac5cc Follow-up for 180065ebb0 - removal of redundant parentheses 2016-12-15 02:34:02 +02:00
Igor Babaev
5cf6fd3e39 Adjusted test results after merge. 2016-12-14 12:11:02 -08:00
Igor Babaev
441fa0056d Fixed bug mdev-11488.
The patch for bug mdev-10882 tried to fix it by providing an
implementation of the virtual method build_clone for the class
Item_cache. It's turned out that it is not easy provide a valid
implementation for Item_cache::build_clone(). At the same time
if the condition that can be pushed into a materialized view
contains a cached item this item can be substituted for a basic
constant of the same value. In such a way we can avoid building
proper clones for Item_cache objects when constructing pushdown
conditions.
2016-12-14 10:13:52 -08:00
Daniel Black
e9ada86265 Travis: add lib{stemmer,xml2,pcre3}-dev 2016-12-14 10:54:03 +04:00
Sergei Golubchik
44e799590e buildbot issues
* rpm upgrade fix
* update test results
* valgrind warning
2016-12-13 21:42:30 +01:00
Marko Mäkelä
65b4d7457e Merge the test innodb.innodb_misc1 into innodb.innodb. 2016-12-13 11:52:23 +02:00
Alexey Botchkov
d26b9f670d MDEV-11470 JSON_KEYS accepts arguments in invalid format.
Now JSON functions return warnings if arguments are invalid.
2016-12-13 12:39:48 +04:00
Sergei Golubchik
1b7a794b73 MDEV-11540 Unexpected system threads in the process list
name innodb background threads as such
2016-12-12 22:33:27 +01:00
Sergei Golubchik
85416269c3 MDEV-11518 Assertion `!table || (!table->read_set || bitmap_is_set(table->read_set, field_index))' failed in Field_long::val_int()
QUICK_RANGE_SELECT::init_ror_merged_scan() should
also set and restore TABLE::vcol_set
2016-12-12 20:44:46 +01:00
Sergei Golubchik
c697ddc315 cleanup: remove unused handler table flag 2016-12-12 20:44:46 +01:00
Sergei Golubchik
b3e3356557 bugfix: reset MODE_NO_BACKSLASH_ESCAPES during vcol parsing 2016-12-12 20:44:46 +01:00
Sergei Golubchik
a9a362d3fd Aria: test for ER_KEY_BASED_ON_GENERATED_VIRTUAL_COLUMN
Aria supports virtual columns, but does not support indexes
on virtual columns. Let's issue an appropriate error in this case.
2016-12-12 20:44:46 +01:00
Sergei Golubchik
9d7c3cbe18 CONNECT: simple vcol test 2016-12-12 20:44:45 +01:00
Sergei Golubchik
5d5e83277f weird compilation fix
strangely enough, ?: variant does not link in some older gcc versions:

sql/sql_table.cc:6409: undefined reference to `Alter_inplace_info::ALTER_STORED_GCOL_EXPR'
sql/sql_table.cc:6409: undefined reference to `Alter_inplace_info::ALTER_VIRTUAL_GCOL_EXPR'
2016-12-12 20:44:45 +01:00
Sergei Golubchik
8c876adfb6 Item_func_like: print a not like b instead of !(a like b) 2016-12-12 20:44:45 +01:00
Sergei Golubchik
180065ebb0 Item::print(): remove redundant parentheses
by introducing new Item::precedence() method and using it
to decide whether parentheses are required
2016-12-12 20:44:41 +01:00
Sergei Golubchik
1db438c833 MDEV-11066 use MySQL terminology for "virtual columns" 2016-12-12 20:35:51 +01:00
Sergei Golubchik
6eaa5fd210 bugfix: InnoDB doesn't support ICP on vcols 2016-12-12 20:35:50 +01:00
Sergei Golubchik
44ca4993b8 bugfix: partitioning and keyread on an indexed vcol 2016-12-12 20:35:50 +01:00
Sergei Golubchik
2a0f7a34d6 bugfix: non-deterministic vcols in partitioning 2016-12-12 20:35:50 +01:00
Sergei Golubchik
d1f3763323 bugfix: non-deterministic vcols in indexes 2016-12-12 20:35:49 +01:00
Sergei Golubchik
6b0f4c24ab cleanup: unpack_vcol_info_from_frm
* do setup/cleanup for charset and arena only once per table
  not for every vcol
* don't copy every vcol expression into table's memroot for parsing,
  do it in a temporary String buffer that is freed at the end
* add asserts
2016-12-12 20:35:49 +01:00
Sergei Golubchik
a72f1deb2d rename Virtual_column_info::expr_item
now, when expr_str is gone, expr_item can be unambiguously
renamed to expr.
2016-12-12 20:35:48 +01:00
Sergei Golubchik
a411d7f4f6 store/show vcols as item->print()
otherwise we'd need to store sql_mode *per vcol*
(consider CREATE INDEX...) and how SHOW CREATE TABLE would
support that?

Additionally, get rid of vcol::expr_str, just to make sure
the string is always generated and never leaked in the
original form.
2016-12-12 20:35:41 +01:00
Sergei Golubchik
8b3b6dc377 test how MDL blocks InnoDB purge
it must, because purge opens a table that might be being altered right now
2016-12-12 20:27:43 +01:00
Sergei Golubchik
1cae1af6f9 MDEV-5800 InnoDB support for indexed vcols
* remove old 5.2+ InnoDB support for virtual columns
  * enable corresponding parts of the innodb-5.7 sources
  * copy corresponding test cases from 5.7
  * copy detailed Alter_inplace_info::HA_ALTER_FLAGS flags from 5.7
     - and more detailed detection of changes in fill_alter_inplace_info()
  * more "innodb compatibility hooks" in sql_class.cc to
     - create/destroy/reset a THD (used by background purge threads)
     - find a prelocked table by name
     - open a table (from a background purge thread)

  * different from 5.7:
    - new service thread "thd_destructor_proxy" to make sure all THDs are
      destroyed at the correct point in time during the server shutdown
    - proper opening/closing of tables for vcol evaluations in
       + FK checks (use already opened prelocked tables)
       + purge threads (open the table, MDLock it, add it to tdc, close
         when not needed)
    - cache open tables in vc_templ
    - avoid unnecessary allocations, reuse table->record[0] and table->s->default_values
    - not needed in 5.7, because it overcalculates:
      + tell the server to calculate vcols for an on-going inline ADD INDEX
      + calculate vcols for correct error messages

  * update other engines (mroonga/tokudb) accordingly
2016-12-12 20:27:42 +01:00
Sergei Golubchik
7fca91f2b4 cleanup: InnoDB, dict_create_add_foreign_to_dictionary()
remove 'table' argument, remnant of 5.6, does not exist in 5.7
2016-12-12 20:27:42 +01:00
Sergei Golubchik
528dd5f20c cleanup: InnoDB, remove index_field_t::col_name
* remnant of 5.6, does not exist in 5.7. bad merge?
* also remove dict_table_get_col_name_for_mysql(), it was only
  used when index_field_t::col_name was not NULL
2016-12-12 20:27:41 +01:00
Sergei Golubchik
b66976abb4 cleanup: InnoDB, various minor issues
* fix "unused pending_checkpoint_mutex_key" compiler warning
* clarify/simplify get_field_offset()
* typos, comments
* unused (forgotten) declaration of create_options_are_invalid()
* fix my_error(ER_WRONG_KEY_COLUMN) calls
* crash in row_upd_sec_index_entry()
* double if (ret != 0)
* don't duplucate PSI_INSTRUMENT_ME lines
* useless break; after return();
* remove unused xtradb-only "cursor_read_view" stuff
* code formatting
* simplify dropped column detection
* redundant assignment
2016-12-12 20:27:41 +01:00
Sergei Golubchik
a6f05b9292 cleanup: redundant casts in THD::dec_thread_count
and THD::inc_thread_count
2016-12-12 20:27:41 +01:00
Sergei Golubchik
0d6a773ae1 cleanup: unused handler::check_if_supported_virtual_columns() 2016-12-12 20:27:40 +01:00
Sergei Golubchik
2614a0ab0f extend prelocking to FK-accessed tables 2016-12-12 20:27:40 +01:00
Sergei Golubchik
f136291098 cleanup: sp_head::add_used_tables_to_table_list()
Use TABLE::init_one_table(), don't duplicate it.
Put additional initializations into TABLE::init_one_table_for_prelocking()
2016-12-12 20:27:40 +01:00
Sergei Golubchik
a3614d33e8 cleanup: FOREIGN_KEY_INFO
instead of returning strings for CASCADE/RESTRICT
from every storage engine, use enum values
2016-12-12 20:27:39 +01:00
Sergei Golubchik
94462aa9e6 bugfix: remove broken insert t values () optimization
* wrong results for 5.7 and 10.1- vcols, and for indexed vcols
* only helps in rare cases and only minimally
2016-12-12 20:27:39 +01:00
Sergei Golubchik
56c1f8da4b bugfix: table->get_fields_in_item_tree=true
partitioning was setting subj as a partitioning-specific
hack (GET_FIXED_FIELDS_FLAG field flag to detect partitioning fields)
inside init_lex_with_single_table(). But vcols also use
init_lex_with_single_table(), they run fix_fields()
in open_table_from_share() before partitioning and this messes up
partitioning GET_FIXED_FIELDS_FLAG logic.
2016-12-12 20:27:39 +01:00
Sergei Golubchik
f73bdb685d bugfix: UPDATE and virtual BLOBs
When updating a table with virtual BLOB columns, the following might happen:
- an old record is read from the table, it has no virtual blob values
- update_virtual_fields() is run, vcol blob gets its value into the
  record. But only a pointer to the value is in the table->record[0],
  the value is in Field_blob::value String (but it doesn't have to be!
  it can be in the record, if the column is just a copy of another
  columns: ... b VARCHAR, c BLOB AS (b) ...)
- store_record(table,record[1]), old record now is in record[1]
- fill_record() prepares new values in record[0], vcol blob is updated,
  new value replaces the old one in the Field_blob::value
- now both record[1] and record[0] have a pointer that points to the
  *new* vcol blob value. Or record[1] has a pointer to nowhere if
  Field_blob::value had to realloc.

To resolve this we unlink vcol blobs from the pointer to the
data (in the record[1]). Because the value is not *always* in
the Field_blob::value String, we need to remember what blobs
were unlinked. The orphan memory must be freed manually.

To complicate the matter, ha_update_row() is also used in
multi-update, in REPLACE, in INSERT ... ON DUPLICATE KEY UPDATE,
also on REPLACE ... SELECT, REPLACE DELAYED, and LOAD DATA REPLACE, etc
2016-12-12 20:27:38 +01:00
Sergei Golubchik
aebb1038aa bugfix: multi-UPDATE, vcols, const tables
multi-update was setting up read_set/vcol_set in
multi_update::initialize_tables() that is invoked after
the optimizer (JOIN::optimize_inner()). But some rows - if they're from
const tables - will be read already in the optimizer, and these rows
will not have all necessary column/vcol values.

* multi_update::initialize_tables() uses results from the optimizer
  and cannot be moved to be called earlier.
* multi_update::prepare() is called before the optimizer, but
  it cannot set up read_set/vcol_set, because the optimizer
  might reset them (see SELECT_LEX::update_used_tables()).

As a fix I've added a new method, select_result::prepare_to_read_rows(),
it's called from inside the optimizer just before make_join_statistics().
2016-12-12 20:27:38 +01:00
Sergei Golubchik
0e401bf7bf bugfix: move vcol calculations down into the handler
This fixes a bug where handler::read_range_first (for example)
needed to compare vcol values that were not calculated yet.

As a bonus it fixes few cases where vcols were calculated twice
2016-12-12 20:27:38 +01:00
Sergei Golubchik
b8f51c04d3 bugfix: update-behind-insert
sql_insert.cc calls handler->ha_update_row() for
REPLACE and INSERT... ON DUPLICATE KEY UPDATE
2016-12-12 20:27:37 +01:00
Sergei Golubchik
54ab7db733 cleanup: remove now-unused TABLE::merge_keys 2016-12-12 20:27:37 +01:00
Sergei Golubchik
b38ff28ade bugfix: mark_columns_needed_for_update
cannot use TABLE:merge_keys for that, because Field::part_of_key
was designed to mark fields for KEY_READ, so a field is not a
"part of key", if only prefix of the field is.
2016-12-12 20:27:36 +01:00
Sergei Golubchik
d137b4dbba MDEV-5800 MyISAM support for indexed vcols
* don't issue an error for ER_KEY_BASED_ON_GENERATED_VIRTUAL_COLUMN
* support keyread on vcols
* callback into the server to compute vcol values from mi_check/mi_repair
* DMLs just work. Automatically.
2016-12-12 20:27:36 +01:00
Sergei Golubchik
a418c99200 gcol mysql-test suite from 5.7
update tests and results, fix bugs
2016-12-12 20:27:36 +01:00