Active transactions must not switch table or index definitions on the fly,
for several reasons, including the following:
* copied indexes do not carry any history or locking information;
that is, rollbacks, read views, and record locking would be broken
* huge potential for race conditions, inconsistent reads and writes,
loss of data, and corruption
Instead of trying to track down if the table was changed during a transaction,
acquire appropriate locks that protect the creation and dropping of indexes.
innodb-index.test: Test the locking of CREATE INDEX and DROP INDEX. Test
that consistent reads work across dropped indexes.
lock_rec_insert_check_and_lock(): Relax the lock_table_has() assertion.
When inserting a record into an index, the table must be at least IX-locked.
However, when an index is being created, an IS-lock on the table is
sufficient.
row_merge_lock_table(): Add the parameter enum lock_mode mode, which must
be LOCK_X or LOCK_S.
row_merge_drop_table(): Assert that n_mysql_handles_opened == 0.
Unconditionally drop the table.
ha_innobase::add_index(): Acquire an X or S lock on the table, as appropriate.
After acquiring an X lock, assert that n_mysql_handles_opened == 1.
Remove the comments about dropping tables in the background.
ha_innobase::final_drop_index(): Acquire an X lock on the table.
dict_table_t: Remove version_number, to_be_dropped, and prebuilts.
ins_node_t: Remove table_version_number.
enum lock_mode: Move the definition from lock0lock.h to lock0types.h.
ROW_PREBUILT_OBSOLETE, row_update_prebuilt(), row_prebuilt_table_obsolete():
Remove.
row_prebuilt_t: Remove the declaration from row0types.h.
row_drop_table_for_mysql_no_commit(): Always print a warning if a table
was added to the background drop queue.
kernel_mutex must be released before calling this function.
innobase_mysql_end_print_arbitrary_thd(),
innobase_mysql_prepare_print_arbitrary_thd(): Assert that the
kernel_mutex is not being held by the current thread.
Non-functional change:
Move the prototypes of
innobase_mysql_prepare_print_arbitrary_thd() and
innobase_mysql_end_print_arbitrary_thd() from lock0lock.c to
ha_prototypes.h
Suggested by: Marko
Approved by: Marko
allocating compressed page frames or their control blocks. Also note
that if buf_buddy_alloc() is used for allocating a control block,
it must be initialized before releasing buf_pool->mutex.
buf_page_init_for_read(): When the page hash check fails after
buf_buddy_alloc(), free the uninitialized control block before freeing
the compressed page frame. This fixes a potential error in
buf_buddy_relocate_block().
mutex is temporarily released.
buf_LRU_free_block(), buf_buddy_alloc_clean(): Add an output parameter that
will be assigned TRUE when the buffer pool mutex is released.
This bug was spotted by and fix provided by Sunny.
value. Document this change in behaviour, and make all callers invoke
the function right after dtuple_create().
dict_create_sys_fields_tuple(): Add a missing "break" statement to the loop
that checks if there are any column prefixes in the index.
row_get_prebuilt_insert_row(): Do not set the fields to the SQL NULL value,
now that dict_table_copy_types() takes care of it.
enough prefixes of externally stored columns, so that purge will not have
to dereference any BLOB pointers, which may be invalid. This will not be
necessary for logging inserts, because inserts are no-ops in purge, and
the record will remain locked during transaction rollback.
TODO: in dict_build_table_def_step() or dict_build_index_def_step(),
prevent the creation of tables with too many columns for which a
prefix index is defined. This is because there is a size limit of undo
log records, and for each prefix-indexed column, the log must store
REC_MAX_INDEX_COL_LEN + BTR_EXTERN_FIELD_REF_SIZE bytes.
trx_undo_page_report_insert(): Assert that the index is clustered.
trx_undo_page_fetch_ext(): New function, for fetching the BLOB prefix
in trx_undo_page_report_modify().
trx_undo_page_report_modify(): Write long enough prefixes of the externally
stored columns to the undo log.
trx_undo_rec_get_partial_row(): Remove the parameter "ext". Assert that
the undo log contains long enough prefixes of the externally stored columns.
purge_node_t: Remove the field "ext".
when row_build() was changed to prefetch all externally stored column
prefixes that occur in ordering fields of an index.
row_build(): Add the parameter col_table for determining which
externally stored columns need to be fetched.
row_merge_read_clustered_index(): Pass new_table as the said parameter,
so that newly added indexes containing column prefix indexes of externally
stored columns will work.
* Change terminology:
wait lock -> requested lock
waited lock -> blocking lock
new: requesting transaction (the trx what owns the requested lock)
new: blocking transaction (the trx that owns the blocking lock)
* Add transaction ids to INFORMATION_SCHEMA.INNODB_LOCK_WAITS. This is
somewhat redundant because transaction ids can be found in INNODB_LOCKS
(which can be joined with INNODB_LOCK_WAITS) but would help users to
write shorter joins (one table less) in some cases where they want to
find which transaction is blocking which.
Suggested by: Ken
Approved by: Heikki
Only add indexed BLOBs to row_ext.
trx_undo_rec_get_partial_row(): Move the BLOB fetching to row_ext_create().
row_build(): Pass only those BLOBs to row_ext_create() that are referenced by
ordering columns of some indexes, similar to trx_undo_rec_get_partial_row().
row_ext_create(): Add the parameter "tuple". Move the implementation
from row0ext.ic to row0ext.c.
row_ext_lookup_ith(), row_ext_lookup(): Return a const pointer. Remove
the parameters "field" and "f_len". Make the row_ext_t* parameter const.
row_ext_t: Remove the field zip_size.
field_ref_zero[]: Declare in btr0types.h instead of btr0cur.h.
row_ext_lookup_low(): Rename to row_ext_cache_fill() and change the
signature.
univ.i: Do not define UNIV_DEBUG, UNIV_ZIP_DEBUG.
btr_cur_del_unmark_for_ibuf(): Use the same comment in both btr0cur.c and
btr0cur.h. Wrap long lines.
contents end up with conflicting versions of a record's state. The zipped
page record was not being marked as "(un)deleted" because we were not
passing the zipped page contents to the (un)delete function, which first
(un)delete marks the uncompressed version and then based on whether
page_zip is NULL or not (un)delete marks the record in the compressed page.
for dropping the index trees, and set the dictionary operation flag, similar
to what ha_innobase::add_index() does. This should ensure correct crash
recovery.
only for those externally stored columns that occur in the ordering columns
of indexes. Prefetch the prefixes of those columns, because the clustered
index record and the BLOBs may have been deleted by the time when the
purge thread needs to read the BLOB prefixes.
row_ext_create(): Add the debug assertion ut_ad(ut_is_2pow(zip_size)).
buf_page_get_release_on_io(): Removed this unused function.
ibuf_build_entry_from_ibuf_rec(): Justify why it is not necessary to
add system columns to the dummy table pointed to by the dummy secondary index.
page_zip_rec_set_deleted(): Add a page_zip_validate() assertion.
inserted, uncommitted clustered index records when determining if a
secondary index record that contains a column prefix of an externally
stored column is referencing the clustered index record.
field_ref_zero[]: A BLOB pointer full of zero, for use in comparisons.
btr_copy_externally_stored_field_prefix(): Assert that the BLOB pointer is set.
row_ext_lookup_ith(), row_ext_lookup(), row_ext_lookup_low(): Document
that field_ref_zero is returned when the BLOB cannot be fetched.
row_ext_lookup_low(): Return field_ref_zero and *len = 0 when the
BLOB pointer is unset.
row_build_index_entry(): Return NULL when a needed BLOB pointer cannot
be dereferenced (row_ext_lookup returns field_ref_zero). Check the
return value for NULL in callers.
row_vers_impl_x_locked_off_kernel(): Avoid comparisons when
row_build_index_entry() returns NULL.
row_vers_old_has_index_entry(): Ignore records for which
row_build_index_entry() returns NULL. The entry should never be NULL
in rollback, but it may be NULL in purge.
row_merge_buf_add(): Assert that row_ext_lookup() does not return
field_ref_zero. The table will be locked during index creation.
column prefix of an externally stored column.
row_upd_ext_fetch(): New function.
row_upd_index_replace_new_col_vals(),
row_upd_index_replace_new_col_vals_index_pos(): Fetch prefixes of
externally stored columns when they are needed for column prefix
indexes. For memory allocation, add the parameter ext_heap. Avoid
repeating the inner loop after finding a matching upd_field->field_no.
innodb_information_schema.test. Add tests that display most columns from
INFORMATION_SCHEMA.INNODB_LOCKS. Test that quoting of table names works
and respects SQL_MODE='ANSI_QUOTES'.
innobase_print_identifier(): Remove.
innobase_convert_identifier(): New function,
based on innobase_print_identifier().
innobase_convert_name(): New function, similar to ut_print_namel(), but
using a memory buffer.
ut_print_namel(): Use innobase_convert_name().
fill_innodb_locks_from_cache(): Convert lock_table and lock_index by
calling innobase_convert_name().
Implement a limit on the memory used by the INNODB_TRX, INNODB_LOCKS and
INNODB_LOCK_WAITS tables. The maximum allowed memory is defined with the
macro TRX_I_S_MEM_LIMIT.
Approved by: Marko (via IM)
Add the query in information_schema.innodb_trx.trx_query. Add it even
though it is available in information_schema.processlist.info to make
inconsistencies between those two tables obvious.
It is rather confusting to see a transaction shown in innodb_trx and
innodb_locks that holds a lock on one table and the corresponding query
in processlist executing INSERT on another table. We do not want users
to contact us asking to explain that. It is caused by the fact that the
data for innodb_* tables and processlist is fetched at different time.
Approved by: Marko
page_zip_get_trailer_len(), page_zip_write_header(): Correct the
UNIV_MEM_ASSERT_RW() assertions.
page_zip_validate(): Read the validity bits of page, page_zip, and
page_zip->data.
Introduce a generic soultion to the common problem that MySQL do not add
functions needed by us in a reasonable time.
Start with a function that retrieves THD::thread_id, this is needed for
the information_schema.innodb_trx.mysql_thread_id column.
Approved by: Marko
a compressed table in the system tablespace.
db0err.h: Introduce the error code DB_TABLE_ZIP_NO_IBD. Replace the
#define directives with an enum, to ease future code merges. These
error codes are never written out to files or displayed to the user.
Thus they need not remain constant.
dict_build_table_def_step(): Return DB_TABLE_ZIP_NO_IBD instead of DB_ERROR.
create_table_def(): Report ER_ILLEGAL_HA_CREATE_OPTION "KEY_BLOCK_SIZE"
when the table creation fails with DB_TABLE_ZIP_NO_IBD.
defined in innodb_redefine.h must not be undefined. After this change,
innodb_redefine.h will define mutex0_free instead of mutex_free, and
everything is fine.
redefined so that the dynamic plugin can replace the builtin InnoDB
in MySQL 5.1.
ha_innodb.cc, handler0alter.cc: #include "univ.i" before any other InnoDB
header files or before defining any symbols
innodb_redefine.h: New file, to contain a mapping of symbols. The idea
is that this file will be replaced in the build process; because this
is a large file that can be generated automatically, it does not make sense
to keep it under version control.
univ.i: #include "innodb_redefine.h" and #define ha_innobase ha_innodb
Makefile.am (ha_innodb_la_CXXFLAGS): Remove -Dha_innobase=ha_innodb
NOTE: there are still some issues in the source code. One known issue is
the #undef mutex_free in sync0sync.h, which will cause the plugin to call the
function mutex_free in the builtin InnoDB. The preprocessor symbols defined
in innodb_redefine.h must not be undefined or redefined anywhere in the code.
page_zip_write_rec(), page_zip_write_blob_ptr(), page_zip_write_node_ptr(),
page_zip_write_trx_id_and_roll_ptr(), page_zip_clear_rec(),
page_zip_rec_set_deleted(), page_zip_rec_set_owned(), page_zip_dir_insert(),
page_zip_dir_delete(), page_zip_dir_add_slot(), page_zip_reorganize(),
page_zip_copy(), page_zip_get_trailer_len(), page_zip_write_header():
Assert that the complete contents of the compressed page is defined.
buf_LRU_free_block(): Check that the block descriptor contains valid data.
buf_buddy_relocate(): Check that the source block contains valid data.
buf_page_get_gen(): Do not dereference bpage after calling buf_relocate().
This avoids a bogus Valgrind warning; the memory itself was valid.
buf_page_hash_get(): Check that the returned block descriptor contains
valid data.
recovered transactions from new ones. Until r1594, they were distinguished
by trx->sess == NULL.
trx_t: Add the bitfield is_recovered.
trx_lists_init_at_db_start(): Set trx->is_recovered.
trx_create(): Initialize trx->is_recovered = 0.
trx_print(): Display information about trx->is_recovered.
trx_rollback_or_clean_all_without_sess(): Skip new transactions.
Protect all accesses of trx_sys->trx_list with kernel_mutex.
trx_roll_crash_recv_trx, trx_roll_max_undo_no, trx_roll_progress_printed_pct:
Made these variables static.
Add innodb_locks.lock_data column and some relevant tests.
For record locks this column represents the ordering fields of the
locked row in a human readable, SQL-valid, format.
Approved by: Marko
on mispredicted branches: opcontrol --event=BR_CND_MISSP_EXEC:45000
UT_SORT_FUNCTION_BODY(): Copy the array with memcpy(3).
cmp_dtuple_rec_with_match(), cmp_rec_rec_with_match(),
cmp_debug_dtuple_rec_with_match(): Add UNIV_UNLIKELY hints around
tests for REC_INFO_MIN_REC_FLAG.
Change mach_read_int_type() to return an integer type (ullint) instead
of array of bytes that later needs to be converted to an appropriate
integer type.
Approved by: Sunny
log_block_calc_checksum(), log_block_get_checksum(): Add const qualifiers.
log_block_calc_checksum(): Split a long expression to more readable chunks.
Some compilers might optimize this better.
