Commit graph

727 commits

Author SHA1 Message Date
marko
4a1d105a31 branches/zip: Merge 1830:1862 from trunk. 2007-09-20 08:36:22 +00:00
marko
81332ff9d0 branches/zip: Improve Valgrind instrumentation in the merge block I/O
of fast index creation.
2007-09-19 13:52:43 +00:00
marko
8b4d78802d branches/zip: Fix some bugs in file-based merge sort.
row_merge_read_rec(): Correct a typo in a comment.  Fix error in
arithmetics when the record spans two blocks.

row_merge_write_rec_low(): Add a "size" parameter.  Add debug assertions
about extra_size and size.

row_merge_write_rec(): After writing a record, properly advance the
buffer pointer.
2007-09-19 11:40:51 +00:00
marko
ec4059555b branches/zip: cmp_rec_rec_simple(): Compare n_uniq columns instead of
all columns present in offsets.  Add a const qualifier to the
dict_index_t* parameter.
2007-09-19 11:31:00 +00:00
marko
23c5767ec6 branches/zip: dict_index_get_n_unique_in_tree(), dict_index_get_n_unique():
Add const qualifiers.
2007-09-19 09:49:58 +00:00
marko
f6aec10b49 branches/zip: innodb-index.test: Add more columns and rows, to ensure
that row_merge_blocks() will have some work to do when
row_merge_block_t is shrunk to 8192 bytes.

Currently, this will cause a debug assertion failure, because
row_merge_cmp() is considering all columns, not just the unique ones.
2007-09-19 09:37:09 +00:00
marko
0d9050315e branches/zip: rec_init_offsets(): Fix a bug that was introduced in r1518:
Correctly handle node pointer records containing variable-length columns
with two-byte length.
2007-09-19 09:05:16 +00:00
marko
fdcf0b1fcd branches/zip: row_merge_buf_add(): Correct the processing of
column prefixes in clustered indexes.

innodb-index.test: Add test cases for creating primary keys
containing column prefixes.
2007-09-19 07:14:36 +00:00
marko
98d2ca10b5 branches/zip: row_ext_lookup_ith(): New function, sliced from
row_ext_lookup().
2007-09-19 07:11:13 +00:00
marko
5de8acee56 branches/zip: ha_innobase::add_index(): Translate DB_TOO_BIG_RECORD
via my_error().  Add a test case to innodb-index.test.
2007-09-17 10:43:51 +00:00
marko
62ede4fcd4 branches/zip: innodb-index.test: Make the column b much longer, to hit
the 65,535-byte VARCHAR length limit of MySQL.
2007-09-17 08:42:10 +00:00
marko
21d303fed2 branches/zip: Move the code related to fast index creation
(smart ALTER TABLE) from ha_innodb.cc to a separate module, handler0alter.cc.
2007-09-13 09:31:54 +00:00
marko
b60b6d142d branches/zip: innodb-index.test: Remove a bogus comment that should have
been removed in r1808.
2007-09-12 19:13:18 +00:00
marko
448568332d branches/zip: ha_node_t: Enclose the field buf_block_t* block inside
#ifdef UNIV_DEBUG.  This should have been done in r1014.
2007-09-12 10:44:53 +00:00
marko
746c3316e1 branches/zip: row_merge_create_temporary_table(): On error, free the
data dictionary object new_table.  This fixes a memory leak.
2007-09-12 07:27:42 +00:00
marko
e9276f80bc branches/zip: ha_innobase::add_index(): Fix a race condition that
existed before the fast-index-creation code was merged, in r1414.

Do not call row_prebuilt_table_obsolete(innodb_table) until after
the table has been renamed to a temporary name.  Otherwise, the
following could happen when creating a clustered index on table T:

(a1) copy T to T#1
(a2) mark T obsolete
(b1) add the name of T to the background drop list
(a3) rename T to T#2 and T#1 to T
(b2) drop the table on the background

These steps are executed by two threads, a and b.

If b1 is executed after a3, all is fine: the name T#2 will be added
to the background drop list.  If b1 is executed before a3, the name
T will be added, and T#2 will survive.
2007-09-11 06:20:19 +00:00
marko
5f3faa0991 branches/zip: Minor cleanup.
row_add_table_to_background_drop_list(): Replace the dict_table_t*
parameter with const char* (the table name).

row_check_table_for_mysql(): Replace prebuilt->table with table.
2007-09-11 05:09:19 +00:00
inaam
40381aed20 branches/zip: Undo changes commited in r1832.
These should have gone to branches/fts.

Spotted by: Marko and Ken
2007-09-11 00:41:02 +00:00
inaam
04fb05b13e branches/zip: undo changes made in r1763.
These should have gone in branches/fts

Spotted by: Marko and Ken
2007-09-11 00:38:13 +00:00
inaam
538665bc2e branches/zip: Enabled group commit functionality with XA.
Reviewed by: Heikki
2007-09-10 17:06:44 +00:00
marko
57afe1d5d2 branches/zip: Merge 1809:1830 from trunk. 2007-09-10 10:09:01 +00:00
marko
313dabb919 branches/zip: UNIV_MEM_ASSERT_AND_FREE():
Use UNIV_MEM_ASSERT_W() instead of UNIV_MEM_ASSERT_RW().
The memory area need not be initialized.
This mistake was made in r1815.
2007-09-07 12:39:36 +00:00
marko
45b97e559f branches/zip: Decode table names when displaying error messages in
fast index creation.

innobase_convert_tablename(): New function, to remove the filename
encoding of a database and table name.

ha_innobase::add_index(): Invoke innobase_convert_tablename() before
calling my_error().

innodb-index.result: Update the error messages.
2007-09-07 09:05:52 +00:00
marko
c10625308b branches/zip: ha_innobase::add_index(): Allocate a new transaction for
the dictionary operations.

TODO: ha_innobase::final_drop_index() should do the same, and set
trx->dict_operation as well.
2007-09-06 12:43:29 +00:00
marko
aebefbe1e8 branches/zip: mem_init_buf(), mem_erase_buf(): Add UNIV_MEM_ASSERT_W(). 2007-09-06 12:25:38 +00:00
marko
f449682428 branches/zip: univ.i: Remove the duplicate UNIV_MEM_DEBUG that was
accidentally added in r1815.
2007-09-06 11:44:00 +00:00
marko
ddc116e14a branches/zip: Improve Valgrind instrumentation.
UNIV_MEM_ASSERT_RW(): New macro, to check that the contents of a memory
area is defined.

UNIV_MEM_ASSERT_W(): New macro, to check that a memory area is writable.

UNIV_MEM_ASSERT_AND_FREE(): New macro, to check that the memory is
writable before declaring it free (unwritable).  This replaces UNIV_MEM_FREE()
in many places.

mem_init_buf(): Declare the memory undefined.

mem_erase_buf(): Declare the memory freed.
2007-09-06 11:36:35 +00:00
marko
462c911fe8 branches/zip: Merge 1792:1809 from trunk. 2007-09-06 07:12:12 +00:00
marko
ffe9a4c52e branches/zip: ha_innobase:add_index(): Report HA_ERR_TABLE_EXIST when the
temporary table cannot be created or renamed.

innodb-index.test,result: Add test cases where a temporary table already
exists.
2007-09-05 11:27:41 +00:00
marko
84ff578142 branches/zip: Simplify crash recovery in fast index creation.
trx_t: Remove dict_undo_list and dict_redo_list.

innobase_create_temporary_tablename(): Replace TEMP_TABLE_PREFIX with
a table name suffix "#1" or "#2".  In this way, the user can restore
precious data, should anything go wrong.  It is possible to reach an
inconsistent state, because the creation, deletion and renaming of
single-table tablespaces are not transactional.

ut_print_namel(), fil_make_ibd_name(), innobase_rename_table(): Remove
the special treatment of TEMP_TABLE_PREFIX.

Introduce TEMP_INDEX_PREFIX == 0xff for temporary indexes.  This byte
cannot occur in index names since MySQL 4.1.  However, it might have
been possible to use this byte in MySQL 4.0.

recv_recovery_from_checkpoint_finish(): Call the new function
row_merge_drop_temp_indexes(), to drop all indexes whose name starts
with the byte 0xff.

row_merge_rename_indexes(): Renamed from row_merge_rename_index().
Remove the parameter "index".

row_drop_table_for_mysql(): Unconditionally call trx_commit_for_mysql().

row_drop_table_for_mysql_no_commit(): Correct the function commit,
based on the corrected comment of row_drop_table_for_mysql().  Rely on
table->to_be_dropped instead of TEMP_TABLE_PREFIX.

ha_innobase::add_index(): Simplify the control flow.
2007-09-05 10:18:03 +00:00
marko
44aa4febc9 branches/zip: lock_rec_enqueue_waiting(): When a record lock wait occurs
in a dictionary operation, display also the name of the index.
2007-09-05 08:57:59 +00:00
vasil
ccd409b569 lock_get_type() needs to be accessed outside of the lock module, but it
is inline and accesses lock_t members.

As advised by Marko rename lock_get_type() to lock_get_type_low() and
create a public non-inline function lock_get_type() which calls
lock_get_type_low().

Approved by:	Marko
2007-09-04 07:54:29 +00:00
vasil
de61e2c143 Use MAP_ANON if MAP_ANONYMOUS is not defined, if neither of both is
defined, fall back to using malloc/free.

Approved by:	Marko
2007-09-04 07:29:59 +00:00
vasil
8393a88e43 Add functions that retrieve some lock_t characteristics, needed in
INFORMATION_SCHEMA implementation:

lock_get_trx_id()
lock_get_mode_str()
lock_get_type_str()
lock_get_table_id()
lock_get_table_name()
lock_rec_get_index_name()
lock_rec_get_space_id()
lock_rec_get_page_no()

Approved by:	Marko
2007-09-03 12:16:11 +00:00
marko
9d65ac4ca4 branches/zip: row_merge_drop_table(): Replace the
row_drop_table_for_mysql() call with a call to
row_drop_table_for_mysql_no_commit().  The last parameter of
the function is ibool drop_db, not ibool do_commit.  Also,
since r1790 of trunk it is not necessary to copy table->name.
2007-09-03 10:34:25 +00:00
vasil
1e3bf9ef79 branches/zip:
Move lock_rec_find_set_bit() from lock0priv.h to lock0lock.h. This
function is needed in the INFORMATION_SCHEMA implementation which
currently resides in trx/

Approved by:	Marko
2007-09-03 09:08:25 +00:00
marko
140a706079 branches/zip: Merge 1783:1792 from trunk. 2007-09-03 08:44:39 +00:00
marko
a2943ec761 branches/zip: Enclose the definition of dict_index_t::trx_id in
#ifdef ROW_MERGE_IS_INDEX_USABLE, to avoid growing the memory
footprint until MySQL implements and calls the virtual method
handler::is_index_available() or something equivalent.
2007-09-03 07:13:59 +00:00
marko
c056d2c3f5 branches/zip: Introduce TEMP_INDEX_PREFIX == 0xff for indexes being populated
in fast index creation.  The prefix was previously TEMP_TABLE_PREFIX == '/',
which may occur in index names.

ut_print_namel(): Do not assume that all '/' are separators between
database and table names.
2007-08-31 09:58:22 +00:00
marko
31d63cd08f branches/zip: Merge 1664:1783 from trunk. 2007-08-30 09:21:25 +00:00
marko
94eee9360d branches/zip: Add const qualifiers to some XID* parameters. 2007-08-30 06:43:27 +00:00
marko
095630bfed branches/zip: Relax the string type check of some built-in functions
of the InnoDB SQL parser.

pars_is_string_type(): New function.  Returns TRUE iff mtype equals
DATA_VARCHAR, DATA_CHAR, DATA_FIXBINARY, or DATA_BINARY.

pars_resolve_func_data_type(): Replace checks for DATA_VARCHAR with
pars_is_string_type().  For example, the NAME column of SYS_INDEXES,
SYS_TABLES, and SYS_COLUMNS is of DATA_BINARY type, although it
contains a UTF-8 character string.  This affects the functions
TO_NUMBER, BINARY_TO_NUMBER, LENGTH, INSTR, and SUBSTR.
2007-08-29 09:54:11 +00:00
marko
666c63777c branches/zip: pars_func_get_class(), pars_resolve_func_data_type():
Replace if-else with switch-case.
2007-08-29 09:40:18 +00:00
marko
8462fba615 branches/zip: ha_innobase::add_index(): Remove a remark about row0mysql.c
in a comment about latching the data dictionary.  The data dictionary may
be latched elsewhere too.
2007-08-29 08:53:29 +00:00
marko
c13f191357 branches/zip: Remove some checks if an index or a table is a temporary one
created in fast index creation.

dict_load_indexes(): Always complain if the first index is not clustered.

lock_table_enqueue_waiting(): Always complain about lock waits in
a dictionary operation.

row_merge_rename_tables(): Add an assertion that dict_sys->mutex is
being held.

row_undo_mod_del_unmark_sec_and_undo_update(): Make the test about
temporary indexes more readable.

row_create_table_for_mysql(): Do not retry creating a temporary table
in fast index creation.  Orphaned temporary tables will have to be dropped
in crash recovery.
2007-08-29 08:45:34 +00:00
marko
9129593724 branches/zip: Simplify the locking of the data dictionary.
row_upd_index_is_referenced(), row_upd_check_references_constraints(),
row_purge_parse_undo_rec(): Remove duplicated code that calls
row_mysql_unfreeze_data_dictionary().

row_undo_ins_remove_clust_rec():
Assert that the data dictionary is X-latched.  Do not attempt to re-latch
the data dictionary, because doing so introduces race conditions.  This
re-latching was probably made unnecessary already in r1676, which keeps
the data dictionary X-latched in ha_innobase::add_index() for the complete
duration of the data dictionary operations.

row_undo_ins_parse_undo_rec(): Do not attempt to acquire the data dictionary
latch.  The data dictionary should be already X-latched when a transaction
that modifies system tables is rolled back.  This also holds in the rollback
during crash recovery: trx_rollback_active() will X-latch the data dictionary
when needed.
2007-08-29 06:36:10 +00:00
marko
63583d80a7 branches/zip: hash_create(): Remove two unused variables. They should have
been removed in r1752.
2007-08-29 05:48:37 +00:00
inaam
98c1aefe7c branches/zip: The freed_page_clock for both buf_pool and block starts ticking
only after the buffer_cache has warmed up. During the initial warmup
phase these values are set to zero. We should recommend to move the
block to the start of the LRU in this case.

Also added some comments to buf0buf.c about the LRU logic.

reviewed by: Sunny
2007-08-27 19:41:56 +00:00
vasil
9584b46995 branches/zip: Add trx_get_id() and trx_get_que_state_str() functions to
retrieve trx_t's properties that will be exported in INFORMATION_SCHEMA tables.

Approved by:	Marko
2007-08-23 09:29:28 +00:00
vasil
617cbacd07 branches/zip: Use hash_table_clear() to initialize the array instead of
for()-loop.

Suggested by:	Sunny
Approved by:	Marko
2007-08-23 07:01:24 +00:00