mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-01-18 04:53:01 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
mariadb/sql/sql_class.h

2394 lines
76 KiB
C
Raw Normal View History

my_strtoll10-x86.s: Corrected spelling in copyright text Makefile.am: Don't update the files from BitKeeper Many files: Removed "MySQL Finland AB & TCX DataKonsult AB" from copyright header Adjusted year(s) in copyright header Many files: Added GPL copyright text Removed files: Docs/Support/colspec-fix.pl Docs/Support/docbook-fixup.pl Docs/Support/docbook-prefix.pl Docs/Support/docbook-split Docs/Support/make-docbook Docs/Support/make-makefile Docs/Support/test-make-manual Docs/Support/test-make-manual-de Docs/Support/xwf
2006-12-31 01:02:27 +01:00
/* Copyright (C) 2000-2006 MySQL AB
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC)
2001-12-06 13:10:51 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
Many files: Changed header to GPL version 2 only
2006-12-23 20:17:15 +01:00
the Free Software Foundation; version 2 of the License.
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC)
2001-12-06 13:10:51 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC)
2001-12-06 13:10:51 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
/* Classes in mysql */
BUG#10241 cygwin port: invalid pragma interface directives - Introduce ifdefs so we can control when to use #pragma interface on cygwin
2005-05-04 15:05:56 +02:00
#ifdef USE_PRAGMA_INTERFACE
Import changeset
2000-07-31 21:29:14 +02:00
#pragma interface /* gcc class implementation */
#endif
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
#include "log.h"
#include "rpl_tblmap.h"
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
BUG#25688 (RBR: circular replication may cause STMT_END_F flags to be skipped): By moving statement end actions from Rows_log_event::do_apply_event() to Rows_log_event::do_update_pos() they will always be executed, even if Rows_log_event::do_apply_event() is skipped because the event originated at the same server. This because Rows_log_event::do_update_pos() is always executed (unless Rows_log_event::do_apply_event() failed with an error, in which case the slave stops with an error anyway). Adding test case. Fixing logic to detect if inside a group. If a rotate event occured when an initial prefix of events for a statement, but for which the table did contain a key, last_event_start_time is set to zero, causing rotate to end the group but without unlocking any tables. This left a lock hanging around, which subsequently triggered an assertion when a second attempt was made to lock the same sequence of tables. In order to solve the above problem, a new flag was added to the relay log info structure that is used to indicate that the replication thread is currently executing a statement. Using this flag, the replication thread is in a group if it is either in a statement or inside a trans- action. The patch also eliminates some gratuitous header file inclusions that were not needed (and caused compile errors) and replaced them with forward definitions.
2007-04-12 08:58:04 +02:00
struct st_relay_log_info;
typedef st_relay_log_info RELAY_LOG_INFO;
Import changeset
2000-07-31 21:29:14 +02:00
class Query_log_event;
class Load_log_event;
Added slave log event to record the master position in of slave thread in slave binlog - needed for synchronization slave during recovery
2001-06-14 22:12:40 +02:00
class Slave_log_event;
Simplistic, experimental framework for Stored Procedures (SPs). Implements creation and dropping of PROCEDUREs, IN, OUT, and INOUT parameters, single-statement procedures, rudimentary multi-statement (begin-end) prodedures (when the client can handle it), and local variables. Missing most of the embedded SQL language, all attributes, FUNCTIONs, error handling, reparses procedures at each call (no caching), etc, etc. Certainly buggy too, but procedures can actually be created and called....
2002-12-08 19:59:22 +01:00
class sp_rcontext;
Code cleanup (and moved sp cache to separate file).
2003-07-03 15:58:37 +02:00
class sp_cache;
manual merge 5.0-runtime -> 5.1->runtime, with 25411 part I
2007-04-26 05:38:12 +02:00
class Lex_input_stream;
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
class Rows_log_event;
Import changeset
2000-07-31 21:29:14 +02:00
ALTER TABLE ... DISABLE KEYS / ALTER TABLE ... ENABLE KEYS
2001-05-11 22:26:12 +02:00
enum enum_enable_or_disable { LEAVE_AS_IS, ENABLE, DISABLE };
Fix for BUG33649: Post-review fixes
2004-05-19 00:18:54 +02:00
enum enum_ha_read_modes { RFIRST, RNEXT, RPREV, RLAST, RKEY, RNEXT_SAME };
Remove DUP_IGNORE from enum_duplicates and instead use a separate ignore flag This allows use to use INSERT IGNORE ... ON DUPLICATE ...
2004-12-31 11:04:35 +01:00
enum enum_duplicates { DUP_ERROR, DUP_REPLACE, DUP_UPDATE };
Changed SQL variable delay_key_write to an enum able delay_key_write as default (as it was in 4.0.2)
2002-08-13 01:18:39 +02:00
enum enum_delay_key_write { DELAY_KEY_WRITE_NONE, DELAY_KEY_WRITE_ON,
DELAY_KEY_WRITE_ALL };
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
enum enum_check_fields
{ CHECK_FIELD_IGNORE, CHECK_FIELD_WARN, CHECK_FIELD_ERROR_FOR_NULL };
enum enum_mark_columns
{ MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE};
Add warnings for single line inserts. To do this I had to convert count_cuted_fields to an enum (to be able to detect setting a NOT NULL field to NULL)
2003-10-11 22:26:39 +02:00
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
extern char internal_table_name[2];
A fix and a test case for Bug#21002 "Derived table not selecting from a "real" table fails in JOINs". This is a regression caused by the fix for Bug 18444. This fix removed the assignment of empty_c_string to table->db performed in add_table_to_list, as neither me nor anyone else knew what it was there for. Now we know it and it's covered with tests: the only case when a table database name can be empty is when the table is a derived table. The fix puts the assignment back but makes it a bit more explicit. Additionally, finally drop sp.result.orig which was checked in by mistake.
2006-07-19 20:33:19 +02:00
extern char empty_c_string[1];
post-review fixes
2004-11-12 13:34:00 +01:00
extern const char **errmesg;
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
#define TC_LOG_PAGE_SIZE 8192
#define TC_LOG_MIN_SIZE (3*TC_LOG_PAGE_SIZE)
#define TC_HEURISTIC_RECOVER_COMMIT 1
#define TC_HEURISTIC_RECOVER_ROLLBACK 2
extern uint tc_heuristic_recover;
'Replicate user variables' task
2003-01-30 18:39:54 +01:00
typedef struct st_user_var_events
{
user_var_entry *user_var_event;
char *value;
ulong length;
Item_result type;
uint charset_number;
} BINLOG_USER_VAR_EVENT;
bug#8646 - deadlock if long transaction causes binlog rotate
2005-02-22 00:15:31 +01:00
#define RP_LOCK_LOG_IS_ALREADY_LOCKED 1
#define RP_FORCE_ROTATE 2
Import changeset
2000-07-31 21:29:14 +02:00
typedef struct st_copy_info {
ha_rows records;
ha_rows deleted;
actions for bug #2709 Affected Rows for ON DUPL. KEY undocumented, perheps illogical 1. added COPY_INFO::updated to work with it in 'insert .. on duplicate' instead of COPY_INFO::deleted 2. added affected rows to output of "info:" in mysqltest.c
2004-03-15 18:36:16 +01:00
ha_rows updated;
Import changeset
2000-07-31 21:29:14 +02:00
ha_rows copied;
Big code cleanup/review before 4.0.2 release. (All commit emails since 4.0.1 checked) This had to be done now, before the 4.1 tree changes to much, to make it easy to propagate bug fixes to the 4.1 tree.
2002-06-11 10:20:31 +02:00
ha_rows error_count;
Import changeset
2000-07-31 21:29:14 +02:00
enum enum_duplicates handle_duplicates;
Big code cleanup/review before 4.0.2 release. (All commit emails since 4.0.1 checked) This had to be done now, before the 4.1 tree changes to much, to make it easy to propagate bug fixes to the 4.1 tree.
2002-06-11 10:20:31 +02:00
int escape_char, last_errno;
Remove DUP_IGNORE from enum_duplicates and instead use a separate ignore flag This allows use to use INSERT IGNORE ... ON DUPLICATE ...
2004-12-31 11:04:35 +01:00
bool ignore;
/* for INSERT ... UPDATE */
INSERT ... ON DUPLICATE KEY UPDATE ...
2002-12-02 20:38:00 +01:00
List<Item> *update_fields;
List<Item> *update_values;
After merge fixes Add support for warnings for prepare of prepared statements Fixed test to work with --ps-protocol Fixed some test results
2005-01-04 12:46:53 +01:00
/* for VIEW ... WITH CHECK OPTION */
post review changes: CHECK OPTION moved to one function view name added to error messages
2004-09-29 15:35:01 +02:00
TABLE_LIST *view;
Import changeset
2000-07-31 21:29:14 +02:00
} COPY_INFO;
class key_part_spec :public Sql_alloc {
public:
const char *field_name;
uint length;
key_part_spec(const char *name,uint len=0) :field_name(name), length(len) {}
Introduce keys in child tables corresponding to FOREIGN KEYs Remove redundant keys in CREATE TABLE and ALTER TABLE
2004-04-21 12:15:43 +02:00
bool operator==(const key_part_spec& other) const;
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
/**
Construct a copy of this key_part_spec. field_name is copied
by-pointer as it is known to never change. At the same time
'length' may be reset in mysql_prepare_create_table, and this
is why we supply it with a copy.
@return If out of memory, 0 is returned and an error is set in
THD.
*/
key_part_spec *clone(MEM_ROOT *mem_root) const
{ return new (mem_root) key_part_spec(*this); }
Import changeset
2000-07-31 21:29:14 +02:00
};
class Alter_drop :public Sql_alloc {
public:
enum drop_type {KEY, COLUMN };
const char *name;
enum drop_type type;
Alter_drop(enum drop_type par_type,const char *par_name)
:name(par_name), type(par_type) {}
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
/**
Used to make a clone of this object for ALTER/CREATE TABLE
@sa comment for key_part_spec::clone
*/
Alter_drop *clone(MEM_ROOT *mem_root) const
{ return new (mem_root) Alter_drop(*this); }
Import changeset
2000-07-31 21:29:14 +02:00
};
class Alter_column :public Sql_alloc {
public:
const char *name;
Item *def;
Alter_column(const char *par_name,Item *literal)
:name(par_name), def(literal) {}
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
/**
Used to make a clone of this object for ALTER/CREATE TABLE
@sa comment for key_part_spec::clone
*/
Alter_column *clone(MEM_ROOT *mem_root) const
{ return new (mem_root) Alter_column(*this); }
Import changeset
2000-07-31 21:29:14 +02:00
};
class Key :public Sql_alloc {
public:
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
enum Keytype { PRIMARY, UNIQUE, MULTIPLE, FULLTEXT, SPATIAL, FOREIGN_KEY};
Import changeset
2000-07-31 21:29:14 +02:00
enum Keytype type;
Cleanups after review of WL#602 Fixed warnings from test suite Some fixes in mysql-test-run script to catch more warnings
2006-05-03 18:40:52 +02:00
KEY_CREATE_INFO key_create_info;
Import changeset
2000-07-31 21:29:14 +02:00
List<key_part_spec> columns;
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
const char *name;
Don't automaticly generate a new key for a foreign key constraint if there is already a usable key. Prefer not automatic keys before automatic keys. If there is two conf
2004-05-11 23:29:52 +02:00
bool generated;
Import changeset
2000-07-31 21:29:14 +02:00
Added support for key_block_size for key and table level (WL#602) Added support for key_block_size to MyISAM. Simplify interface to 'new Key' to make it easier to add new key options. mysqld option --new is used to define where key options are printed. (In 5.3 we should move all key options to after key part definition to avoid problem with reserved names) Fixed some compiler warnings and a memory leak in ssl
2006-05-03 14:59:17 +02:00
Key(enum Keytype type_par, const char *name_arg,
KEY_CREATE_INFO *key_info_arg,
bool generated_arg, List<key_part_spec> &cols)
Cleanups after review of WL#602 Fixed warnings from test suite Some fixes in mysql-test-run script to catch more warnings
2006-05-03 18:40:52 +02:00
:type(type_par), key_create_info(*key_info_arg), columns(cols),
name(name_arg), generated(generated_arg)
Update for BTREE keys in HEAP tables Split handler->option_flag() to handler->table_flags() and handler->index_flags()
2002-04-12 20:35:46 +02:00
{}
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
Key(const Key &rhs, MEM_ROOT *mem_root);
virtual ~Key() {}
Introduce keys in child tables corresponding to FOREIGN KEYs Remove redundant keys in CREATE TABLE and ALTER TABLE
2004-04-21 12:15:43 +02:00
/* Equality comparison of keys (ignoring name) */
Don't automaticly generate a new key for a foreign key constraint if there is already a usable key. Prefer not automatic keys before automatic keys. If there is two conf
2004-05-11 23:29:52 +02:00
friend bool foreign_key_prefix(Key *a, Key *b);
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
/**
Used to make a clone of this object for ALTER/CREATE TABLE
@sa comment for key_part_spec::clone
*/
virtual Key *clone(MEM_ROOT *mem_root) const
{ return new (mem_root) Key(*this, mem_root); }
Import changeset
2000-07-31 21:29:14 +02:00
};
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
class Table_ident;
class foreign_key: public Key {
public:
enum fk_match_opt { FK_MATCH_UNDEF, FK_MATCH_FULL,
FK_MATCH_PARTIAL, FK_MATCH_SIMPLE};
enum fk_option { FK_OPTION_UNDEF, FK_OPTION_RESTRICT, FK_OPTION_CASCADE,
FK_OPTION_SET_NULL, FK_OPTION_NO_ACTION, FK_OPTION_DEFAULT};
Table_ident *ref_table;
List<key_part_spec> ref_columns;
uint delete_opt, update_opt, match_opt;
foreign_key(const char *name_arg, List<key_part_spec> &cols,
Table_ident *table, List<key_part_spec> &ref_cols,
uint delete_opt_arg, uint update_opt_arg, uint match_opt_arg)
Added support for key_block_size for key and table level (WL#602) Added support for key_block_size to MyISAM. Simplify interface to 'new Key' to make it easier to add new key options. mysqld option --new is used to define where key options are printed. (In 5.3 we should move all key options to after key part definition to avoid problem with reserved names) Fixed some compiler warnings and a memory leak in ssl
2006-05-03 14:59:17 +02:00
:Key(FOREIGN_KEY, name_arg, &default_key_create_info, 0, cols),
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
ref_table(table), ref_columns(cols),
delete_opt(delete_opt_arg), update_opt(update_opt_arg),
match_opt(match_opt_arg)
{}
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
foreign_key(const foreign_key &rhs, MEM_ROOT *mem_root);
/**
Used to make a clone of this object for ALTER/CREATE TABLE
@sa comment for key_part_spec::clone
*/
virtual Key *clone(MEM_ROOT *mem_root) const
{ return new (mem_root) foreign_key(*this, mem_root); }
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
};
Import changeset
2000-07-31 21:29:14 +02:00
typedef struct st_mysql_lock
{
TABLE **table;
uint table_count,lock_count;
THR_LOCK_DATA **locks;
} MYSQL_LOCK;
class LEX_COLUMN : public Sql_alloc
{
public:
String column;
uint rights;
LEX_COLUMN (const String& x,const uint& y ): column (x),rights (y) {}
};
#include "sql_lex.h" /* Must be here */
Follow the coding style with class names.
2007-05-10 16:27:36 +02:00
class Delayed_insert;
new EXPLAIN fixed bug in mysql-test/create-test-result fixed bug in union-subselect engine
2002-09-26 22:08:22 +02:00
class select_result;
Fix the tree: plugin-related changes, necessary for CSV storage engine to be statically compiled. Derived from antony's patch.
2006-01-19 22:40:56 +01:00
class Time_zone;
Import changeset
2000-07-31 21:29:14 +02:00
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
#define THD_SENTRY_MAGIC 0xfeedd1ff
#define THD_SENTRY_GONE 0xdeadbeef
#define THD_CHECK_SENTRY(thd) DBUG_ASSERT(thd->dbug_sentry == THD_SENTRY_MAGIC)
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
struct system_variables
{
WL#2936 "Server Variables for Plugins" Implement support for plugins to declare server variables. Demonstrate functionality by removing InnoDB specific code from sql/* New feature for HASH - HASH_UNIQUE flag New feature for DYNAMIC_ARRAY - initializer accepts preallocated ptr. Completed support for plugin reference counting.
2007-03-02 17:43:45 +01:00
/*
How dynamically allocated system variables are handled:
The global_system_variables and max_system_variables are "authoritative"
They both should have the same 'version' and 'size'.
When attempting to access a dynamic variable, if the session version
is out of date, then the session version is updated and realloced if
neccessary and bytes copied from global to make up for missing data.
*/
ulong dynamic_variables_version;
char* dynamic_variables_ptr;
uint dynamic_variables_head; /* largest valid variable offset */
uint dynamic_variables_size; /* how many bytes are in use */
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulonglong myisam_max_extra_sort_file_size;
ulonglong myisam_max_sort_file_size;
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
ulonglong max_heap_table_size;
ulonglong tmp_table_size;
Changed thd variables max_join_size and select_limit to type ha_rows. This fixed some optimization problems when using -DBIG_TABLES Portabilty fixes for OpenUnix and HPUX Added C and C++ version numbers to mysqlbug
2002-12-20 13:58:27 +01:00
ha_rows select_limit;
ha_rows max_join_size;
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
ulong auto_increment_increment, auto_increment_offset;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong bulk_insert_buff_size;
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
ulong join_buff_size;
ulong long_query_time;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong max_allowed_packet;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
ulong max_error_count;
Many files: Added the MAX_LENGTH_FOR_SORT_DATA option filesort.cc: Added the MAX_LENGTH_FOR_SORT_DATA option
2003-04-24 13:33:33 +02:00
ulong max_length_for_sort_data;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
ulong max_sort_length;
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
ulong max_tmp_tables;
Fixed Bug#2491, INSERT DELAYED causes binary log unusable by mysqlbinlog (replication is ok) Setting insert delayed threads = 0 at mysql_init_command.
2004-03-04 18:58:36 +01:00
ulong max_insert_delayed_threads;
WL#2126 - Multi_read_range. Added the required structures and functions for handing over multiple key ranges to the table handler.
2004-12-23 21:45:10 +01:00
ulong multi_range_count;
parallel-repair available in mysqld
2003-05-04 18:43:37 +02:00
ulong myisam_repair_threads;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong myisam_sort_buff_size;
BUG#12232: Add a server option to treat NULL values as equal when calculating MyISAM index statistics (like 4.0 did) (patch #3, with review #1 & #2 feedback addressed)
2005-09-21 00:18:29 +02:00
ulong myisam_stats_method;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong net_buffer_length;
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
ulong net_interactive_timeout;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong net_read_timeout;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
ulong net_retry_count;
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
ulong net_wait_timeout;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong net_write_timeout;
Better names for two system variables: - heuristic => optimizer_prune_level - plan_search_depth => optimizer_search_depth
2004-05-20 16:47:43 +02:00
ulong optimizer_prune_level;
ulong optimizer_search_depth;
Many files: New feature: preload indexes into key cache. mi_preload.c: new file Many files: Added preload statement.
2003-06-12 13:29:02 +02:00
ulong preload_buff_size;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong query_cache_type;
ulong read_buff_size;
ulong read_rnd_buff_size;
A lot of fixes to Precision math Mostly about precision/decimals of the results of the operations
2005-05-05 17:06:49 +02:00
ulong div_precincrement;
Added interface for first set of dynamic variables.
2002-06-28 18:30:09 +02:00
ulong sortbuff_size;
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
ulong thread_handling;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
ulong tx_isolation;
Merge
2005-02-11 22:33:52 +01:00
ulong completion_type;
merge with 4.0 for more memory allocation variables.
2003-10-13 14:50:30 +02:00
/* Determines which non-standard SQL behaviour should be enabled */
Added support sql_mode, which can be used to produce various outputs of SHOW CREATE TABLE 'name'. Depending on the mode, the output can be compatible with various databases, including earlier versions of MySQL .
2003-01-16 01:04:50 +01:00
ulong sql_mode;
Recursion support made for SP (BUG#10100).
2005-11-23 00:11:19 +01:00
ulong max_sp_recursion_depth;
VIEW two TABLE_LIST copy eliminated
2004-07-16 00:15:55 +02:00
/* check of key presence in updatable view */
sql_updatable_view_key replaced with updatable_views_with_limit now default behaviour do not prevent any updates of view. Also updating VIEWs of queries without LIMIT clause will not be checked at all (there will not be even warning) (changes according to discussion on last dev-conf)
2004-10-07 11:13:42 +02:00
ulong updatable_views_with_limit;
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
ulong default_week_format;
Added thread variable max_seeks_for_key Change optimizer to prefer key lookups before table scan Change table scans to be done after tables with constrains on scanned table
2003-06-27 02:04:54 +02:00
ulong max_seeks_for_key;
Add new user variables for tuning memory usage: query_alloc_block_size, query_prealloc_size, range_alloc_block_size,transaction_alloc_block_size and transaction_prealloc_size Add more checks for "out of memory" detection in range optimization
2003-10-11 21:00:24 +02:00
ulong range_alloc_block_size;
ulong query_alloc_block_size;
ulong query_prealloc_size;
ulong trans_alloc_block_size;
ulong trans_prealloc_size;
Changed --log-warnings to be integer instead of boolean. Given --skip-log-warnings will disable warnings, --log-warnings will increment warning level by one, or the level can be given as an optional argument. Default level is 1. Changed aborted connection warning to be logged only if the level is > 1.
2004-06-01 16:29:24 +02:00
ulong log_warnings;
This is full commit of group_concat with support subselects
2003-03-18 00:07:40 +01:00
ulong group_concat_max_len;
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
ulong ndb_autoincrement_prefetch_sz;
ulong ndb_index_stat_cache_entries;
ulong ndb_index_stat_update_freq;
ulong binlog_format; // binlog format for this thd (see enum_binlog_format)
processing thread specific queries
2002-12-29 22:46:48 +01:00
/*
In slave thread we need to know in behalf of which
thread the query is being run to replicate temp tables properly
*/
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
my_thread_id pseudo_thread_id;
processing thread specific queries
2002-12-29 22:46:48 +01:00
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
my_bool low_priority_updates;
my_bool new_mode;
WL#3527: Extend IGNORE INDEX so places where index is ignored can be specified Currently MySQL allows one to specify what indexes to ignore during join optimization. The scope of the current USE/FORCE/IGNORE INDEX statement is only the FROM clause, while all other clauses are not affected. However, in certain cases, the optimizer may incorrectly choose an index for sorting and/or grouping, and produce an inefficient query plan. This task provides the means to specify what indexes are ignored/used for what operation in a more fine-grained manner, thus making it possible to manually force a better plan. We do this by extending the current IGNORE/USE/FORCE INDEX syntax to: IGNORE/USE/FORCE INDEX [FOR {JOIN | ORDER | GROUP BY}] so that: - if no FOR is specified, the index hint will apply everywhere. - if MySQL is started with the compatibility option --old_mode then an index hint without a FOR clause works as in 5.0 (i.e, the index will only be ignored for JOINs, but can still be used to compute ORDER BY). See the WL#3527 for further details.
2007-03-05 18:08:41 +01:00
/*
compatibility option:
- index usage hints (USE INDEX without a FOR clause) behave as in 5.0
*/
my_bool old_mode;
invalidation of locking for write tables (BUG#2693) fixed linking query_prealloc_size to query cache presence
2004-03-04 17:32:55 +01:00
my_bool query_cache_wlock_invalidate;
Condition pushdown v.2
2005-02-11 22:05:24 +01:00
my_bool engine_condition_pushdown;
Enabled usage of new system variables for ndb - ndb_use_exact_count - ndb_autoincrement_prefetch_sz - ndb_use_transactions - ndb_force_send moved "inlined" functions to .cc file since they are virtual anyways enabled printout od ndb errors in warnings even if mapping existst to mysql error code
2004-11-17 09:15:53 +01:00
my_bool ndb_force_send;
Fixed incorrect handling of renamed fields, forcing copy of table if needed and added optional system variable ndb_use_copying_alter_table
2006-06-13 16:44:30 +02:00
my_bool ndb_use_copying_alter_table;
Enabled usage of new system variables for ndb - ndb_use_exact_count - ndb_autoincrement_prefetch_sz - ndb_use_transactions - ndb_force_send moved "inlined" functions to .cc file since they are virtual anyways enabled printout od ndb errors in warnings even if mapping existst to mysql error code
2004-11-17 09:15:53 +01:00
my_bool ndb_use_exact_count;
my_bool ndb_use_transactions;
ndb - wl#2624 re-commit due to bk problem
2005-09-15 02:33:28 +02:00
my_bool ndb_index_stat_enable;
WL#2977 and WL#2712 global and session-level variable to set the binlog format (row/statement), and new binlog format called "mixed" (which is statement-based except if only row-based is correct, in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release): SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default; the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha. It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below). The added tests test the possibility or impossibility to SET, their effects, and the mixed mode, including in prepared statements and in stored procedures and functions. Caveats: a) The mixed mode will not work for stored functions: in mixed mode, a stored function will always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()). b) for the same reason, changing the thread's binlog format inside a stored function is refused with an error message. c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask Dmitri). Additionally, as the binlog format is now changeable by each user for his session, I remove the implication which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1 (not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled phantom protection). Plus fixes for compiler warnings.
2006-02-25 22:21:03 +01:00
Faster alter table code for 5.1.
2005-07-22 22:43:59 +02:00
my_bool old_alter_table;
Preliminary support for options --secure-auth, --old-passwords Support for option --old-protocol was removed. Some test performed. Tests for SSL and replication are pending. More strict following to specification for --old-passwords is in the TODO.
2003-07-08 00:36:14 +02:00
my_bool old_passwords;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
WL#2936 "Server Variables for Plugins" Implement support for plugins to declare server variables. Demonstrate functionality by removing InnoDB specific code from sql/* New feature for HASH - HASH_UNIQUE flag New feature for DYNAMIC_ARRAY - initializer accepts preallocated ptr. Completed support for plugin reference counting.
2007-03-02 17:43:45 +01:00
plugin_ref table_plugin;
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
New SQL variables "collation_server" and "collation_database"
2003-09-15 13:31:04 +02:00
/* Only charset part of these variables is sensible */
loaddata.result, loaddata.test: Adding test case. sql_yacc.yy: Adding TEXT_STRING_filesystem, which converts from character_set_client to character_set_conversion. Replacing TEXT_STRING_sys to TEXT_STRING_filesystem in LOAD DATA and SELECT INTO OUTFILE contexts. sql_class.h, sql_class.cc: Adding character_set_filesystem variable, and charset_is_character_set_filesystem flag (to avoid conversion when it's not necessary). set_var.h, set_var.cc: Adding sys_var_character_set_filesystem mysqld.cc: Adding --character-set-filesystem startup option.
2006-01-18 09:55:38 +01:00
CHARSET_INFO *character_set_filesystem;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
CHARSET_INFO *character_set_client;
Some SHOW VARIABLES have been renamed: collation_client -> character_set_client collation_results -> character_set_results character_set -> character_set_server SET NAMES now doesn't start client->server conversion SET CHARACTER SET now starts both client->server and server->client conversion
2003-05-21 14:44:12 +02:00
CHARSET_INFO *character_set_results;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
New SQL variables "collation_server" and "collation_database"
2003-09-15 13:31:04 +02:00
/* Both charset and collation parts of these variables are important */
CHARSET_INFO *collation_server;
CHARSET_INFO *collation_database;
Variable names have been renamed: client_collation -> collation_client result_collation -> collation_results connection_collation -> collation_connection
2003-04-23 15:19:22 +02:00
CHARSET_INFO *collation_connection;
Simplified 'wrong xxx name' error messages by introducing 'general' ER_WRONG_NAME error Cleaned up (and disabled part of) date/time/datetime format patch. One can't anymore change default read/write date/time/formats. This is becasue the non standard datetime formats can't be compared as strings and MySQL does still a lot of datetime comparisons as strings Changed flag argument to str_to_TIME() and get_date() from bool to uint Removed THD from str_to_xxxx functions and Item class. Fixed core dump when doing --print-defaults Move some common string functions to strfunc.cc Dates as strings are now of type my_charset_bin instead of default_charset() Introduce IDENT_QUOTED to not have to create an extra copy of simple identifiers (all chars < 128) Removed xxx_FORMAT_TYPE enums and replaced them with the old TIMESTAMP_xxx enums Renamed some TIMESTAMP_xxx enums to more appropriate names Use defines instead of integers for date/time/datetime string lengths Added to build system and use the new my_strtoll10() function.
2003-11-03 13:01:59 +01:00
WL#2928 Date Translation NRE (implemented by by Josh Chamas)
2006-07-04 14:40:40 +02:00
/* Locale Support */
MY_LOCALE *lc_time_names;
WL#1264 "Per-thread time zone support infrastructure". Added basic per-thread time zone functionality (based on public domain elsie-code). Now user can select current time zone (from the list of time zones described in system tables). All NOW-like functions honor this time zone, values of TIMESTAMP type are interpreted as values in this time zone, so now our TIMESTAMP type behaves similar to Oracle's TIMESTAMP WITH LOCAL TIME ZONE (or proper PostgresSQL type). WL#1266 "CONVERT_TZ() - basic time with time zone conversion function". Fixed problems described in Bug #2336 (Different number of warnings when inserting bad datetime as string or as number). This required reworking of datetime realted warning hadling (they now generated at Field object level not in conversion functions). Optimization: Now Field class descendants use table->in_use member instead of current_thd macro.
2004-06-18 08:11:31 +02:00
Time_zone *time_zone;
Removed not used define YY_MAGIC_BELOW Made year 2000 handling more uniform Removed year 2000 handling out from calc_days() The above removes some bugs in date/datetimes with year between 0 and 200 Now we get a note when we insert a datetime value into a date column For default values to CREATE, don't give errors for warning level NOTE Fixed some compiler failures Added library ws2_32 for windows compilation (needed if we want to compile with IOCP support) Removed duplicate typedef TIME and replaced it with MYSQL_TIME Better (more complete) fix for: Bug#21103 "DATE column not compared as DATE" Fixed properly Bug#18997 "DATE_ADD and DATE_SUB perform year2K autoconversion magic on 4-digit year value" Fixed Bug#23093 "Implicit conversion of 9912101 to date does not match cast(9912101 as date)"
2007-03-23 21:08:31 +01:00
/* DATE, DATETIME and MYSQL_TIME formats */
Simplified 'wrong xxx name' error messages by introducing 'general' ER_WRONG_NAME error Cleaned up (and disabled part of) date/time/datetime format patch. One can't anymore change default read/write date/time/formats. This is becasue the non standard datetime formats can't be compared as strings and MySQL does still a lot of datetime comparisons as strings Changed flag argument to str_to_TIME() and get_date() from bool to uint Removed THD from str_to_xxxx functions and Item class. Fixed core dump when doing --print-defaults Move some common string functions to strfunc.cc Dates as strings are now of type my_charset_bin instead of default_charset() Introduce IDENT_QUOTED to not have to create an extra copy of simple identifiers (all chars < 128) Removed xxx_FORMAT_TYPE enums and replaced them with the old TIMESTAMP_xxx enums Renamed some TIMESTAMP_xxx enums to more appropriate names Use defines instead of integers for date/time/datetime string lengths Added to build system and use the new my_strtoll10() function.
2003-11-03 13:01:59 +01:00
DATE_TIME_FORMAT *date_format;
DATE_TIME_FORMAT *datetime_format;
DATE_TIME_FORMAT *time_format;
BUG#15101 SYSDATE() disregards SET TIMESTAMP. After the ChangeSet 1.1892.20.1 2005/08/24 (Bug #12562) SYSDATE() is not an alias of NOW() and is unsafe for replication. `SYSDATE()' backward compatible aliasing clashes with the idea #12562 fix. To make it safe-replicatable we have to either use RBR or to restore the pre-5.0 style. --sysdate-is-now command line flag was introduced to provide backward compatibility.
2006-03-10 15:47:56 +01:00
my_bool sysdate_is_now;
WL#3527: Extend IGNORE INDEX so places where index is ignored can be specified Currently MySQL allows one to specify what indexes to ignore during join optimization. The scope of the current USE/FORCE/IGNORE INDEX statement is only the FROM clause, while all other clauses are not affected. However, in certain cases, the optimizer may incorrectly choose an index for sorting and/or grouping, and produce an inefficient query plan. This task provides the means to specify what indexes are ignored/used for what operation in a more fine-grained manner, thus making it possible to manually force a better plan. We do this by extending the current IGNORE/USE/FORCE INDEX syntax to: IGNORE/USE/FORCE INDEX [FOR {JOIN | ORDER | GROUP BY}] so that: - if no FOR is specified, the index hint will apply everywhere. - if MySQL is started with the compatibility option --old_mode then an index hint without a FOR clause works as in 5.0 (i.e, the index will only be ignored for JOINs, but can still be used to compute ORDER BY). See the WL#3527 for further details.
2007-03-05 18:08:41 +01:00
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
};
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
/* per thread status variables */
typedef struct system_status_var
{
ulong bytes_received;
ulong bytes_sent;
ulong com_other;
ulong com_stat[(uint) SQLCOM_END];
ulong created_tmp_disk_tables;
ulong created_tmp_tables;
ulong ha_commit_count;
ulong ha_delete_count;
ulong ha_read_first_count;
ulong ha_read_last_count;
ulong ha_read_key_count;
ulong ha_read_next_count;
ulong ha_read_prev_count;
ulong ha_read_rnd_count;
ulong ha_read_rnd_next_count;
ulong ha_rollback_count;
ulong ha_update_count;
ulong ha_write_count;
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
ulong ha_prepare_count;
ulong ha_discover_count;
ulong ha_savepoint_count;
ulong ha_savepoint_rollback_count;
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
/* KEY_CACHE parts. These are copies of the original */
ulong key_blocks_changed;
ulong key_blocks_used;
ulong key_cache_r_requests;
ulong key_cache_read;
ulong key_cache_w_requests;
ulong key_cache_write;
/* END OF KEY_CACHE parts */
ulong net_big_packet_count;
ulong opened_tables;
Table definition cache, part 2 The table opening process now works the following way: - Create common TABLE_SHARE object - Read the .frm file and unpack it into the TABLE_SHARE object - Create a TABLE object based on the information in the TABLE_SHARE object and open a handler to the table object Other noteworthy changes: - In TABLE_SHARE the most common strings are now LEX_STRING's - Better error message when table is not found - Variable table_cache is now renamed 'table_open_cache' - New variable 'table_definition_cache' that is the number of table defintions that will be cached - strxnmov() calls are now fixed to avoid overflows - strxnmov() will now always add one end \0 to result - engine objects are now created with a TABLE_SHARE object instead of a TABLE object. - After creating a field object one must call field->init(table) before using it - For a busy system this change will give you: - Less memory usage for table object - Faster opening of tables (if it's has been in use or is in table definition cache) - Allow you to cache many table definitions objects - Faster drop of table
2005-11-23 21:45:02 +01:00
ulong opened_shares;
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
ulong select_full_join_count;
ulong select_full_range_join_count;
ulong select_range_count;
ulong select_range_check_count;
ulong select_scan_count;
ulong long_query_count;
ulong filesort_merge_passes;
ulong filesort_range_count;
ulong filesort_rows;
ulong filesort_scan_count;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
/* Prepared statements and binary protocol */
Manual merge (4.1->5.0)
2005-06-16 22:34:35 +02:00
ulong com_stmt_prepare;
ulong com_stmt_execute;
ulong com_stmt_send_long_data;
ulong com_stmt_fetch;
ulong com_stmt_reset;
ulong com_stmt_close;
Change Last_query_cost status variable from global to thread-local.
2005-05-18 05:39:10 +02:00
Improve comments around FLUSH STATUS It's not possible to flush the global status variables in 5.0 Update test case so it works by recording the value of handle_rollback before and compare it to the value after
2006-10-30 13:35:57 +01:00
/*
Status variables which it does not make sense to add to
global status variable counter
*/
Change Last_query_cost status variable from global to thread-local.
2005-05-18 05:39:10 +02:00
double last_query_cost;
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
} STATUS_VAR;
/*
Improve comments around FLUSH STATUS It's not possible to flush the global status variables in 5.0 Update test case so it works by recording the value of handle_rollback before and compare it to the value after
2006-10-30 13:35:57 +01:00
This is used for 'SHOW STATUS'. It must be updated to the last ulong
variable in system_status_var which is makes sens to add to the global
counter
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
*/
Post-merge fixes.
2005-06-16 23:58:36 +02:00
#define last_system_status_var com_stmt_close
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
Fix the tree: plugin-related changes, necessary for CSV storage engine to be statically compiled. Derived from antony's patch.
2006-01-19 22:40:56 +01:00
#ifdef MYSQL_SERVER
fixed subselects with temporary tables (SCRUM) fixed memory leacks
2003-01-25 01:25:52 +01:00
void free_tmp_table(THD *thd, TABLE *entry);
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
Fixes during review of new code - Mostly indentation fixes - Added missing test - Ensure that Item_func_case() checks for stack overruns - Use real_item() instead of (Item_ref*) item - Fixed wrong error handling
2005-07-04 02:42:33 +02:00
/* The following macro is to make init of Query_arena simpler */
#ifndef DBUG_OFF
#define INIT_ARENA_DBUG_INFO is_backup_arena= 0
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
#else
#define INIT_ARENA_DBUG_INFO
Fixes during review of new code - Mostly indentation fixes - Added missing test - Ensure that Item_func_case() checks for stack overruns - Use real_item() instead of (Item_ref*) item - Fixed wrong error handling
2005-07-04 02:42:33 +02:00
#endif
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
class Query_arena
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
{
public:
/*
List of items created in the parser for this query. Every item puts
itself to the list on creation (see Item::Item() for details))
*/
Item *free_list;
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
MEM_ROOT *mem_root; // Pointer to current memroot
Fixes after merge with 4.1 FOUND is not a reserved keyword anymore Added Item_field::set_no_const_sub() to be able to mark fields that can't be substituted Added 'simple_select' method to be able to quickly determinate if a select_result is a normal SELECT Note that the 5.0 tree is not yet up to date: Sanja will have to fix multi-update-locks for this merge to be complete
2004-11-03 11:39:38 +01:00
#ifndef DBUG_OFF
- implement inheritance of sp_instr: public Query_arena. We need every instruction to have its own arena, because we want to track instruction's state (INITIALIZED_FOR_SP -> EXECUTED). Because of `if' statements and other conditional instructions used in stored procedures, not every instruction of a stored procedure gets executed during the first (or even subsequent) execution of the procedure. So it's better if we track the execution state of every instruction independently. All instructions of a given procedure now also share sp_head's mem_root, but keep their own free_list. This simplifies juggling with free Item lists in sp_head::execute. - free_items() moved to be a member of Query_arena. - logic of 'backup_arena' debug member of Query_arena has been changed to support multi-backups. Until now, TRUE 'backup_arena' meant that there is exactly one active backup of the THD arena. Now it means simply that the arena is used for backup, so that we can't accidentally overwrite an existing backup. This allows doing multiple backups, e.g. in sp_head::execute and Cursor::fetch, when THD arena is already backed up but we want to set yet another arena (usually the 'permanent' arena, to save permanent transformations/optimizations of a parsed tree).
2005-06-23 18:22:08 +02:00
bool is_backup_arena; /* True if this arena is used for backup. */
Fixes after merge with 4.1 FOUND is not a reserved keyword anymore Added Item_field::set_no_const_sub() to be able to mark fields that can't be substituted Added 'simple_select' method to be able to quickly determinate if a select_result is a normal SELECT Note that the 5.0 tree is not yet up to date: Sanja will have to fix multi-update-locks for this merge to be complete
2004-11-03 11:39:38 +01:00
#endif
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
enum enum_state
Fix to compile with msvc: converted static const int Item_arena::* to enum members, undefine ERROR
2004-08-24 12:44:15 +02:00
{
Patch two (the final one) for Bug#7306 "the server side preparedStatement error for LIMIT placeholder". The patch adds grammar support for LIMIT ?, ? and changes the type of ST_SELECT_LEX::select_limit,offset_limit from ha_rows to Item*, so that it can point to Item_param.
2005-06-07 12:11:36 +02:00
INITIALIZED= 0, INITIALIZED_FOR_SP= 1, PREPARED= 2,
CONVENTIONAL_EXECUTION= 3, EXECUTED= 4, ERROR= -1
Fix to compile with msvc: converted static const int Item_arena::* to enum members, undefine ERROR
2004-08-24 12:44:15 +02:00
};
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
Change Item_arena::state to enum
2004-08-31 12:07:02 +02:00
enum_state state;
Fix for bug#4912 "mysqld crashs in case a statement is executed a second time". The bug was caused by incompatibility of negations elimination algorithm and PS: during first statement execute a subtree with negation was replaced with equivalent subtree without NOTs. The problem was that although this transformation was permanent, items of the new subtree were created in execute-local memory. The patch adds means to check if it is the first execute of a prepared statement, and if this is the case, to allocate items in memory of the prepared statement. The implementation: - backports Item_arena from 5.0 - adds Item_arena::is_stmt_prepare(), Item_arena::is_first_stmt_execute(). - deletes THD::allocate_temporary_pool_for_ps_preparing(), THD::free_temporary_pool_for_ps_preparing(); they were redundant. and adds a few invariants: - thd->free_list never contains junk (= freed items) - thd->current_arena is never null. If there is no prepared statement, it points at the thd. The rest of the patch contains mainly mechanical changes and cleanups.
2004-08-21 00:02:46 +02:00
/* We build without RTTI, so dynamic_cast can't be used. */
enum Type
{
STATEMENT, PREPARED_STATEMENT, STORED_PROCEDURE
};
Adjust to the changed Query_arena constructor: main_mem_root is moved out of class Query_arena.
2005-06-22 09:59:13 +02:00
Query_arena(MEM_ROOT *mem_root_arg, enum enum_state state_arg) :
free_list(0), mem_root(mem_root_arg), state(state_arg)
- implement inheritance of sp_instr: public Query_arena. We need every instruction to have its own arena, because we want to track instruction's state (INITIALIZED_FOR_SP -> EXECUTED). Because of `if' statements and other conditional instructions used in stored procedures, not every instruction of a stored procedure gets executed during the first (or even subsequent) execution of the procedure. So it's better if we track the execution state of every instruction independently. All instructions of a given procedure now also share sp_head's mem_root, but keep their own free_list. This simplifies juggling with free Item lists in sp_head::execute. - free_items() moved to be a member of Query_arena. - logic of 'backup_arena' debug member of Query_arena has been changed to support multi-backups. Until now, TRUE 'backup_arena' meant that there is exactly one active backup of the THD arena. Now it means simply that the arena is used for backup, so that we can't accidentally overwrite an existing backup. This allows doing multiple backups, e.g. in sp_head::execute and Cursor::fetch, when THD arena is already backed up but we want to set yet another arena (usually the 'permanent' arena, to save permanent transformations/optimizations of a parsed tree).
2005-06-23 18:22:08 +02:00
{ INIT_ARENA_DBUG_INFO; }
Implementation of Monty's idea about clear_alloc_root() optimization and cleanup of work with memory roots in THD/Statement/Item_arena. Added assertions preventing memory allocation on bzero'ed MEM_ROOT since it is worked by pure luck and was very ineffective.
2004-09-23 11:48:17 +02:00
/*
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
This constructor is used only when Query_arena is created as
backup storage for another instance of Query_arena.
Implementation of Monty's idea about clear_alloc_root() optimization and cleanup of work with memory roots in THD/Statement/Item_arena. Added assertions preventing memory allocation on bzero'ed MEM_ROOT since it is worked by pure luck and was very ineffective.
2004-09-23 11:48:17 +02:00
*/
- implement inheritance of sp_instr: public Query_arena. We need every instruction to have its own arena, because we want to track instruction's state (INITIALIZED_FOR_SP -> EXECUTED). Because of `if' statements and other conditional instructions used in stored procedures, not every instruction of a stored procedure gets executed during the first (or even subsequent) execution of the procedure. So it's better if we track the execution state of every instruction independently. All instructions of a given procedure now also share sp_head's mem_root, but keep their own free_list. This simplifies juggling with free Item lists in sp_head::execute. - free_items() moved to be a member of Query_arena. - logic of 'backup_arena' debug member of Query_arena has been changed to support multi-backups. Until now, TRUE 'backup_arena' meant that there is exactly one active backup of the THD arena. Now it means simply that the arena is used for backup, so that we can't accidentally overwrite an existing backup. This allows doing multiple backups, e.g. in sp_head::execute and Cursor::fetch, when THD arena is already backed up but we want to set yet another arena (usually the 'permanent' arena, to save permanent transformations/optimizations of a parsed tree).
2005-06-23 18:22:08 +02:00
Query_arena() { INIT_ARENA_DBUG_INFO; }
Fix for bug#4912 "mysqld crashs in case a statement is executed a second time". The bug was caused by incompatibility of negations elimination algorithm and PS: during first statement execute a subtree with negation was replaced with equivalent subtree without NOTs. The problem was that although this transformation was permanent, items of the new subtree were created in execute-local memory. The patch adds means to check if it is the first execute of a prepared statement, and if this is the case, to allocate items in memory of the prepared statement. The implementation: - backports Item_arena from 5.0 - adds Item_arena::is_stmt_prepare(), Item_arena::is_first_stmt_execute(). - deletes THD::allocate_temporary_pool_for_ps_preparing(), THD::free_temporary_pool_for_ps_preparing(); they were redundant. and adds a few invariants: - thd->free_list never contains junk (= freed items) - thd->current_arena is never null. If there is no prepared statement, it points at the thd. The rest of the patch contains mainly mechanical changes and cleanups.
2004-08-21 00:02:46 +02:00
virtual Type type() const;
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
virtual ~Query_arena() {};
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
Patch two (the final one) for Bug#7306 "the server side preparedStatement error for LIMIT placeholder". The patch adds grammar support for LIMIT ?, ? and changes the type of ST_SELECT_LEX::select_limit,offset_limit from ha_rows to Item*, so that it can point to Item_param.
2005-06-07 12:11:36 +02:00
inline bool is_stmt_prepare() const { return state == INITIALIZED; }
A preparatory patch to help adding JOIN::transform() and move one-time query transformations to the PREPARE stage (prepared statements).
2005-06-21 19:15:21 +02:00
inline bool is_first_sp_execute() const
{ return state == INITIALIZED_FOR_SP; }
Give Item_arena::is_stmt_prepare a more descriptive name (it marks the code that is active for SP as well in 5.0)
2005-06-02 22:02:47 +02:00
inline bool is_stmt_prepare_or_first_sp_execute() const
{ return (int)state < (int)PREPARED; }
Change Item_arena::state to enum
2004-08-31 12:07:02 +02:00
inline bool is_first_stmt_execute() const { return state == PREPARED; }
A fix and test case for Bug#6050 "EXECUTE stmt reports ambiguous field names with ident. tables fr. diff. schemata": revise all uses of Item_field and make them prepared-statements friendly when necessary.
2004-10-22 12:47:35 +02:00
inline bool is_stmt_execute() const
{ return state == PREPARED || state == EXECUTED; }
Fixes after merge with 4.1 FOUND is not a reserved keyword anymore Added Item_field::set_no_const_sub() to be able to mark fields that can't be substituted Added 'simple_select' method to be able to quickly determinate if a select_result is a normal SELECT Note that the 5.0 tree is not yet up to date: Sanja will have to fix multi-update-locks for this merge to be complete
2004-11-03 11:39:38 +01:00
inline bool is_conventional() const
post merge fixes
2004-09-10 01:22:44 +02:00
{ return state == CONVENTIONAL_EXECUTION; }
Adjust to the changed Query_arena constructor: main_mem_root is moved out of class Query_arena.
2005-06-22 09:59:13 +02:00
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline void* alloc(size_t size) { return alloc_root(mem_root,size); }
inline void* calloc(size_t size)
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
void *ptr;
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
if ((ptr=alloc_root(mem_root,size)))
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
bzero(ptr, size);
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
return ptr;
}
inline char *strdup(const char *str)
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
{ return strdup_root(mem_root,str); }
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline char *strmake(const char *str, size_t size)
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
{ return strmake_root(mem_root,str,size); }
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline bool LEX_STRING_make(LEX_STRING *lex_str, const char *str,
size_t size)
Changed several char* to LEX_STRING*.
2006-10-16 18:57:33 +02:00
{
return ((lex_str->str=
strmake_root(mem_root, str, (lex_str->length= size)))) == 0;
}
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline void *memdup(const void *str, size_t size)
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
{ return memdup_root(mem_root,str,size); }
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline void *memdup_w_gap(const void *str, size_t size, uint gap)
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
void *ptr;
if ((ptr= alloc_root(mem_root,size+gap)))
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
memcpy(ptr,str,size);
return ptr;
}
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
void set_query_arena(Query_arena *set);
- implement inheritance of sp_instr: public Query_arena. We need every instruction to have its own arena, because we want to track instruction's state (INITIALIZED_FOR_SP -> EXECUTED). Because of `if' statements and other conditional instructions used in stored procedures, not every instruction of a stored procedure gets executed during the first (or even subsequent) execution of the procedure. So it's better if we track the execution state of every instruction independently. All instructions of a given procedure now also share sp_head's mem_root, but keep their own free_list. This simplifies juggling with free Item lists in sp_head::execute. - free_items() moved to be a member of Query_arena. - logic of 'backup_arena' debug member of Query_arena has been changed to support multi-backups. Until now, TRUE 'backup_arena' meant that there is exactly one active backup of the THD arena. Now it means simply that the arena is used for backup, so that we can't accidentally overwrite an existing backup. This allows doing multiple backups, e.g. in sp_head::execute and Cursor::fetch, when THD arena is already backed up but we want to set yet another arena (usually the 'permanent' arena, to save permanent transformations/optimizations of a parsed tree).
2005-06-23 18:22:08 +02:00
void free_items();
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
/* Close the active state associated with execution of this statement */
virtual void cleanup_stmt();
PS and SP made compatible in mechanism used for preparing query for rexecutions (Bug #2266)
2004-05-20 01:02:49 +02:00
};
Port of cursors to be pushed into 5.0 tree: - client side part is simple and may be considered stable - server side part now just joggles with THD state to save execution state and has no additional locking wisdom. Lot's of it are to be rewritten.
2004-08-03 12:32:21 +02:00
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
class Server_side_cursor;
Port of cursors to be pushed into 5.0 tree: - client side part is simple and may be considered stable - server side part now just joggles with THD state to save execution state and has no additional locking wisdom. Lot's of it are to be rewritten.
2004-08-03 12:32:21 +02:00
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
@class Statement
@brief State of a single command executed against this connection.
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
One connection can contain a lot of simultaneously running statements,
some of which could be:
- prepared, that is, contain placeholders,
- opened as cursors. We maintain 1 to 1 relationship between
statement and cursor - if user wants to create another cursor for his
query, we create another statement for it.
To perform some action with statement we reset THD part to the state of
that statement, do the action, and then save back modified state from THD
to the statement. It will be changed in near future, and Statement will
be used explicitly.
*/
A fix and a test case for Bug#10760 and complementary cleanups. The idea of the patch is that every cursor gets its own lock id for table level locking. Thus cursors are protected from updates performed within the same connection. Additionally a list of transient (must be closed at commit) cursors is maintained and all transient cursors are closed when necessary. Lastly, this patch adds support for deadlock timeouts to TLL locking when using cursors. + post-review fixes.
2005-07-19 20:21:12 +02:00
class Statement: public ilink, public Query_arena
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
{
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
Statement(const Statement &rhs); /* not implemented: */
Statement &operator=(const Statement &rhs); /* non-copyable */
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
public:
/*
after-review fixes
2003-12-04 20:08:26 +01:00
Uniquely identifies each statement object in thread scope; change during
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
statement lifetime. FIXME: must be const
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
*/
after-review fixes
2003-12-04 20:08:26 +01:00
ulong id;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
/*
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
MARK_COLUMNS_NONE: Means mark_used_colums is not set and no indicator to
handler of fields used is set
MARK_COLUMNS_READ: Means a bit in read set is set to inform handler
that the field is to be read. If field list contains
duplicates, then thd->dup_field is set to point
to the last found duplicate.
MARK_COLUMNS_WRITE: Means a bit is set in write set to inform handler
that it needs to update this field in write_row
and update_row.
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
*/
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
enum enum_mark_columns mark_used_columns;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
LEX_STRING name; /* name for named prepared statements */
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
LEX *lex; // parse tree descriptor
/*
Points to the query associated with this statement. It's const, but
we need to declare it char * because all table handlers are written
in C and need to point to it.
Move Heikki's comment about thd->query comment to proper place (bad merge from 4.0)
2004-10-12 22:17:37 +02:00
Note that (A) if we set query = NULL, we must at the same time set
query_length = 0, and protect the whole operation with the
LOCK_thread_count mutex. And (B) we are ONLY allowed to set query to a
non-NULL value if its previous value is NULL. We do not need to protect
operation (B) with any mutex. To avoid crashes in races, if we do not
know that thd->query cannot change at the moment, one should print
thd->query like this:
(1) reserve the LOCK_thread_count mutex;
(2) check if thd->query is NULL;
(3) if not NULL, then print at most thd->query_length characters from
it. We will see the query_length field as either 0, or the right value
for it.
Assuming that the write and read of an n-bit memory field in an n-bit
computer is atomic, we can avoid races in the above way.
This printing is needed at least in SHOW PROCESSLIST and SHOW INNODB
STATUS.
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
*/
char *query;
uint32 query_length; // current query length
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
Server_side_cursor *cursor;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
public:
Remove THD::stmt_backup
2005-06-22 21:12:25 +02:00
/* This constructor is called for backup statements */
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
Statement() {}
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
Statement(LEX *lex_arg, MEM_ROOT *mem_root_arg,
enum enum_state state_arg, ulong id_arg);
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
virtual ~Statement();
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
/* Assign execution context (note: not all members) of given stmt to self */
void set_statement(Statement *stmt);
Fix for bug#4912 "mysqld crashs in case a statement is executed a second time". The bug was caused by incompatibility of negations elimination algorithm and PS: during first statement execute a subtree with negation was replaced with equivalent subtree without NOTs. The problem was that although this transformation was permanent, items of the new subtree were created in execute-local memory. The patch adds means to check if it is the first execute of a prepared statement, and if this is the case, to allocate items in memory of the prepared statement. The implementation: - backports Item_arena from 5.0 - adds Item_arena::is_stmt_prepare(), Item_arena::is_first_stmt_execute(). - deletes THD::allocate_temporary_pool_for_ps_preparing(), THD::free_temporary_pool_for_ps_preparing(); they were redundant. and adds a few invariants: - thd->free_list never contains junk (= freed items) - thd->current_arena is never null. If there is no prepared statement, it points at the thd. The rest of the patch contains mainly mechanical changes and cleanups.
2004-08-21 00:02:46 +02:00
void set_n_backup_statement(Statement *stmt, Statement *backup);
void restore_backup_statement(Statement *stmt, Statement *backup);
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
/* return class type */
virtual Type type() const;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
};
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
Container for all statements created/used in a connection.
Statements in Statement_map have unique Statement::id (guaranteed by id
assignment in Statement::Statement)
Non-empty statement names are unique too: attempt to insert a new statement
with duplicate name causes older statement to be deleted
A fix and test case for Bug#5315 "mysql_change_user() doesn't free prepared statements."
2004-09-22 13:50:07 +02:00
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
Statements are auto-deleted when they are removed from the map and when the
map is deleted.
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
*/
class Statement_map
{
public:
Statement_map();
A fix and test case for Bug#5315 "mysql_change_user() doesn't free prepared statements."
2004-09-22 13:50:07 +02:00
A fix and a test case for Bug#16365 "Prepared Statements: DoS with too many open statements". The patch adds a new global variable @@max_prepared_stmt_count. This variable limits the total number of prepared statements in the server. The default value of @@max_prepared_stmt_count is 16382. 16382 small statements (a select against 3 tables with GROUP, ORDER and LIMIT) consume 100MB of RAM. Once this limit has been reached, the server will refuse to prepare a new statement and return ER_UNKNOWN_ERROR (unfortunately, we can't add new errors to 4.1 without breaking 5.0). The limit is changeable after startup and can accept any value from 0 to 1 million. In case the new value of the limit is less than the current statement count, no new statements can be added, while the old still can be used. Additionally, the current count of prepared statements is now available through a global read-only variable @@prepared_stmt_count.
2006-04-07 21:37:06 +02:00
int insert(THD *thd, Statement *statement);
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
Statement *find_by_name(LEX_STRING *name)
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
{
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
Statement *stmt;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
stmt= (Statement*)hash_search(&names_hash, (uchar*)name->str,
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
name->length);
return stmt;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
}
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
Statement *find(ulong id)
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
{
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
if (last_found_statement == 0 || id != last_found_statement->id)
WL#1622 "SQL Syntax for Prepared Statements": Post-review fixes (1 of 2)
2004-05-21 02:27:50 +02:00
{
Statement *stmt;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
stmt= (Statement *) hash_search(&st_hash, (uchar *) &id, sizeof(id));
Fix for bug#4236 "Server crash on attempt to execute non-prepared statement": check that statement is not null when accessing it's name.
2004-06-22 09:04:41 +02:00
if (stmt && stmt->name.str)
WL#1622 "SQL Syntax for Prepared Statements": Post-review fixes (1 of 2)
2004-05-21 02:27:50 +02:00
return NULL;
last_found_statement= stmt;
}
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
return last_found_statement;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
}
A fix and a test case for Bug#10760 and complementary cleanups. The idea of the patch is that every cursor gets its own lock id for table level locking. Thus cursors are protected from updates performed within the same connection. Additionally a list of transient (must be closed at commit) cursors is maintained and all transient cursors are closed when necessary. Lastly, this patch adds support for deadlock timeouts to TLL locking when using cursors. + post-review fixes.
2005-07-19 20:21:12 +02:00
/*
Close all cursors of this connection that use tables of a storage
engine that has transaction-specific state and therefore can not
survive COMMIT or ROLLBACK. Currently all but MyISAM cursors are closed.
*/
void close_transient_cursors();
A fix and a test case for Bug#16365 "Prepared Statements: DoS with too many open statements". The patch adds a new global variable @@max_prepared_stmt_count. This variable limits the total number of prepared statements in the server. The default value of @@max_prepared_stmt_count is 16382. 16382 small statements (a select against 3 tables with GROUP, ORDER and LIMIT) consume 100MB of RAM. Once this limit has been reached, the server will refuse to prepare a new statement and return ER_UNKNOWN_ERROR (unfortunately, we can't add new errors to 4.1 without breaking 5.0). The limit is changeable after startup and can accept any value from 0 to 1 million. In case the new value of the limit is less than the current statement count, no new statements can be added, while the old still can be used. Additionally, the current count of prepared statements is now available through a global read-only variable @@prepared_stmt_count.
2006-04-07 21:37:06 +02:00
void erase(Statement *statement);
A fix and test case for Bug#5315 "mysql_change_user() doesn't free prepared statements."
2004-09-22 13:50:07 +02:00
/* Erase all statements (calls Statement destructor) */
A fix and a test case for Bug#16365 "Prepared Statements: DoS with too many open statements". The patch adds a new global variable @@max_prepared_stmt_count. This variable limits the total number of prepared statements in the server. The default value of @@max_prepared_stmt_count is 16382. 16382 small statements (a select against 3 tables with GROUP, ORDER and LIMIT) consume 100MB of RAM. Once this limit has been reached, the server will refuse to prepare a new statement and return ER_UNKNOWN_ERROR (unfortunately, we can't add new errors to 4.1 without breaking 5.0). The limit is changeable after startup and can accept any value from 0 to 1 million. In case the new value of the limit is less than the current statement count, no new statements can be added, while the old still can be used. Additionally, the current count of prepared statements is now available through a global read-only variable @@prepared_stmt_count.
2006-04-07 21:37:06 +02:00
void reset();
~Statement_map();
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
private:
HASH st_hash;
WL#1622 "SQL Syntax for Prepared Statements": post-review fixes: Moved PS name to Statement class, Statement_map now handles name-to-statement resolution. Both named and unnamed statements are now executed in one function (sql_prepare.cc:execute_stmt) Fixed a problem: Malformed sequence of commands from client could cause server to use previously deleted objects. Some code cleanup and small fixes
2004-04-12 23:58:48 +02:00
HASH names_hash;
A fix and a test case for Bug#10760 and complementary cleanups. The idea of the patch is that every cursor gets its own lock id for table level locking. Thus cursors are protected from updates performed within the same connection. Additionally a list of transient (must be closed at commit) cursors is maintained and all transient cursors are closed when necessary. Lastly, this patch adds support for deadlock timeouts to TLL locking when using cursors. + post-review fixes.
2005-07-19 20:21:12 +02:00
I_List<Statement> transient_cursor_list;
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
Statement *last_found_statement;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
};
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
struct st_savepoint {
struct st_savepoint *prev;
char *name;
uint length, nht;
};
enum xa_states {XA_NOTR=0, XA_ACTIVE, XA_IDLE, XA_PREPARED};
query_id and my_xid -> ulonglong fix for binlog+autocommit+tclog comments, style fixes
2005-01-27 22:38:56 +01:00
extern const char *xa_state_names[];
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
typedef struct st_xid_state {
/* For now, this is only used to catch duplicated external xids */
XID xid; // transaction identifier
enum xa_states xa_state; // used by external XA only
bool in_thd;
} XID_STATE;
extern pthread_mutex_t LOCK_xid_cache;
extern HASH xid_cache;
bool xid_cache_init(void);
void xid_cache_free(void);
XID_STATE *xid_cache_search(XID *xid);
bool xid_cache_insert(XID *xid, enum xa_states xa_state);
bool xid_cache_insert(XID_STATE *xid_state);
void xid_cache_delete(XID_STATE *xid_state);
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
@class Security_context
@brief A set of THD members describing the current authenticated user.
*/
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
WL#2787 (Add view definer/owner to the view definition (.frm) to check privileges on used tables and stored routines when using a VIEW.) Part 2 postreview fixes.
2005-09-20 20:20:38 +02:00
class Security_context {
public:
Fixed compiler warnings from gcc 4.0.2: - Added empty constructors and virtual destructors to many classes and structs - Removed some usage of the offsetof() macro to instead use C++ class pointers
2006-02-25 16:46:30 +01:00
Security_context() {} /* Remove gcc warning */
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
/*
host - host of the client
user - user of the client, set to NULL until the user has been read from
the connection
WL#2787 (Add view definer/owner to the view definition (.frm) to check privileges on used tables and stored routines when using a VIEW.) Part 2 postreview fixes.
2005-09-20 20:20:38 +02:00
priv_user - The user privilege we are using. May be "" for anonymous user.
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
ip - client IP
*/
char *host, *user, *priv_user, *ip;
WL#2787 (Add view definer/owner to the view definition (.frm) to check privileges on used tables and stored routines when using a VIEW.) Part 2 postreview fixes.
2005-09-20 20:20:38 +02:00
/* The host privilege we are using */
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
char priv_host[MAX_HOSTNAME];
/* points to host if host is available, otherwise points to ip */
const char *host_or_ip;
ulong master_access; /* Global privileges from mysql.user */
ulong db_access; /* Privileges for current db */
void init();
void destroy();
void skip_grants();
WL#2787 (Add view definer/owner to the view definition (.frm) to check privileges on used tables and stored routines when using a VIEW.) Part 2 postreview fixes.
2005-09-20 20:20:38 +02:00
inline char *priv_host_name()
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
{
return (*priv_host ? priv_host : (char *)"%");
}
fix for bug #17619 Scheduler race conditions - Scheduler is either initialized at server start or never. Starting & stopping is now suspending & resuming. - The scheduler has clear OO interface - Now all calls to the scheduler are synchronous - GLOBAL event_scheduler uses thd::sys_var_tmp (see set_var.cc) - External API is encapsulated into class Events - Includes fixes for all comments of Kostja's review of 19.05.2005 Starting to merge into 5.1-release (5.1.10) and push
2006-05-22 20:46:13 +02:00
bool set_user(char *user_arg);
An attempt to fix a sporadic valgrind memory leak in Event Scheduler: streamline the event worker thread work flow and try to eliminate possibilities for memory corruptions that might have been lurking in previous (complicated) code. This patch: * removes Event_job_data::compile that was never used * cleans up Event_job_data::execute to minimize juggling with thread context and eliminate unneded code paths * Implements Security_context::change/restore_security_context to be able to re-use these methods in all stored programs This is to maybe fix Bug#27733 "Valgrind failures in remove_table_from_cache". Review comments applied.
2007-04-13 22:35:56 +02:00
#ifndef NO_EMBEDDED_ACCESS_CHECKS
bool
change_security_context(THD *thd,
LEX_STRING *definer_user,
LEX_STRING *definer_host,
LEX_STRING *db,
Security_context **backup);
void
restore_security_context(THD *thd, Security_context *backup);
#endif
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
};
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
A registry for item tree transformations performed during
query optimization. We register only those changes which require
a rollback to re-execute a prepared statement or stored procedure
yet another time.
*/
struct Item_change_record;
typedef I_List<Item_change_record> Item_change_list;
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
Better approach for prelocking of tables for stored routines execution and some SP-related cleanups. - We don't have separate stage for calculation of list of tables to be prelocked and doing implicit LOCK/UNLOCK any more. Instead we calculate this list at open_tables() and do implicit LOCK in lock_tables() (and UNLOCK in close_thread_tables()). Also now we support cases when same table (with same alias) is used several times in the same query in SP. - Cleaned up execution of SP. Moved all common code which handles LEX and does preparations before statement execution or complex expression evaluation to auxilary sp_lex_keeper class. Now all statements in SP (and corresponding instructions) that evaluate expression which can contain subquery have their own LEX.
2005-03-04 14:35:28 +01:00
Type of prelocked mode.
See comment for THD::prelocked_mode for complete description.
*/
enum prelocked_mode_type {NON_PRELOCKED= 0, PRELOCKED= 1,
PRELOCKED_UNDER_LOCK_TABLES= 2};
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
Class that holds information about tables which were opened and locked
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
by the thread. It is also used to save/restore this information in
push_open_tables_state()/pop_open_tables_state().
*/
class Open_tables_state
{
public:
/*
open_tables - list of regular tables in use by this thread
temporary_tables - list of temp tables in use by this thread
handler_tables - list of tables that were opened with HANDLER OPEN
and are still in use by this thread
*/
TABLE *open_tables, *temporary_tables, *handler_tables, *derived_tables;
/*
During a MySQL session, one can lock tables in two modes: automatic
or manual. In automatic mode all necessary tables are locked just before
statement execution, and all acquired locks are stored in 'lock'
member. Unlocking takes place automatically as well, when the
statement ends.
Manual mode comes into play when a user issues a 'LOCK TABLES'
statement. In this mode the user can only use the locked tables.
Trying to use any other tables will give an error. The locked tables are
stored in 'locked_tables' member. Manual locking is described in
the 'LOCK_TABLES' chapter of the MySQL manual.
See also lock_tables() for details.
*/
MYSQL_LOCK *lock;
/*
Tables that were locked with explicit or implicit LOCK TABLES.
(Implicit LOCK TABLES happens when we are prelocking tables for
execution of statement which uses stored routines. See description
THD::prelocked_mode for more info.)
*/
MYSQL_LOCK *locked_tables;
Patch to handle some bad situations resulting from the fix for BUG#19995.
2006-06-22 13:28:04 +02:00
/*
CREATE-SELECT keeps an extra lock for the table being
created. This field is used to keep the extra lock available for
lower level routines, which would otherwise miss that lock.
*/
MYSQL_LOCK *extra_lock;
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
/*
prelocked_mode_type enum and prelocked_mode member are used for
indicating whenever "prelocked mode" is on, and what type of
"prelocked mode" is it.
Prelocked mode is used for execution of queries which explicitly
or implicitly (via views or triggers) use functions, thus may need
some additional tables (mentioned in query table list) for their
execution.
First open_tables() call for such query will analyse all functions
used by it and add all additional tables to table its list. It will
also mark this query as requiring prelocking. After that lock_tables()
will issue implicit LOCK TABLES for the whole table list and change
thd::prelocked_mode to non-0. All queries called in functions invoked
by the main query will use prelocked tables. Non-0 prelocked_mode
will also surpress mentioned analysys in those queries thus saving
cycles. Prelocked mode will be turned off once close_thread_tables()
for the main query will be called.
Note: Since not all "tables" present in table list are really locked
thd::prelocked_mode does not imply thd::locked_tables.
*/
prelocked_mode_type prelocked_mode;
ulong version;
uint current_tablenr;
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
enum enum_flags {
BACKUPS_AVAIL = (1U << 0) /* There are backups available */
};
/*
Flags with information about the open tables state.
*/
uint state_flags;
Fix for bug #10055 "Using stored function with information_schema causes empty result set". To enable full access to contents of I_S tables from stored functions or statements that use them, we manipulate with thread's open tables state and ensure that we won't cause deadlock when we open tables by ignoring flushes and name-locks. Building of contents of I_S.TABLES no longer requires locking of tables since we use use handler::info() method with HA_STATUS_AUTO flag instead of handler::update_auto_increment() for obtaining information about auto-increment values. But this also means that handlers have to implement support for HA_STATUS_AUTO flag (particularly InnoDB needs it).
2005-08-08 15:46:06 +02:00
/*
This constructor serves for creation of Open_tables_state instances
which are used as backup storage.
*/
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
Open_tables_state() : state_flags(0U) { }
Fix for bug #10055 "Using stored function with information_schema causes empty result set". To enable full access to contents of I_S tables from stored functions or statements that use them, we manipulate with thread's open tables state and ensure that we won't cause deadlock when we open tables by ignoring flushes and name-locks. Building of contents of I_S.TABLES no longer requires locking of tables since we use use handler::info() method with HA_STATUS_AUTO flag instead of handler::update_auto_increment() for obtaining information about auto-increment values. But this also means that handlers have to implement support for HA_STATUS_AUTO flag (particularly InnoDB needs it).
2005-08-08 15:46:06 +02:00
Open_tables_state(ulong version_arg);
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
void set_open_tables_state(Open_tables_state *state)
{
*this= *state;
}
void reset_open_tables_state()
{
open_tables= temporary_tables= handler_tables= derived_tables= 0;
Patch to handle some bad situations resulting from the fix for BUG#19995.
2006-06-22 13:28:04 +02:00
extra_lock= lock= locked_tables= 0;
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
prelocked_mode= NON_PRELOCKED;
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
state_flags= 0U;
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
}
};
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
@class Sub_statement_state
@brief Used to save context when executing a function or trigger
*/
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
/* Defines used for Sub_statement_state::in_sub_stmt */
#define SUB_STMT_TRIGGER 1
#define SUB_STMT_FUNCTION 2
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
class Sub_statement_state
{
public:
ulonglong options;
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
ulonglong first_successful_insert_id_in_prev_stmt;
ulonglong first_successful_insert_id_in_cur_stmt, insert_id_for_cur_row;
Discrete_interval auto_inc_interval_for_cur_row;
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
ulonglong limit_found_rows;
ha_rows cuted_fields, sent_row_count, examined_row_count;
ulong client_capabilities;
uint in_sub_stmt;
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
bool enable_slow_log;
after-merge fixes.
2006-06-16 11:05:58 +02:00
bool last_insert_id_used;
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
my_bool no_send_ok;
Fix for bug #13825 "Triggers: crash if release savepoint" and for general handling of savepoints in stored routines. Fixed ha_rollback_to_savepoint()/ha_savepoint()/ha_release_savepoint() functions to properly handle savepoints inside of stored functions and triggers. Also now when we invoke stored function or trigger we create new savepoint level. We destroy it at the end of function/trigger execution and return back to old savepoint level.
2005-11-19 13:09:23 +01:00
SAVEPOINT *savepoints;
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
};
fix for bug #17619 Scheduler race conditions - Scheduler is either initialized at server start or never. Starting & stopping is now suspending & resuming. - The scheduler has clear OO interface - Now all calls to the scheduler are synchronous - GLOBAL event_scheduler uses thd::sys_var_tmp (see set_var.cc) - External API is encapsulated into class Events - Includes fixes for all comments of Kostja's review of 19.05.2005 Starting to merge into 5.1-release (5.1.10) and push
2006-05-22 20:46:13 +02:00
/* Flags for the THD::system_thread variable */
enum enum_thread_type
{
NON_SYSTEM_THREAD= 0,
SYSTEM_THREAD_DELAYED_INSERT= 1,
SYSTEM_THREAD_SLAVE_IO= 2,
SYSTEM_THREAD_SLAVE_SQL= 4,
SYSTEM_THREAD_NDBCLUSTER_BINLOG= 8,
SYSTEM_THREAD_EVENT_SCHEDULER= 16,
SYSTEM_THREAD_EVENT_WORKER= 32
};
Manual merge
2007-03-06 21:46:33 +01:00
/**
This class represents the interface for internal error handlers.
Internal error handlers are exception handlers used by the server
implementation.
*/
class Internal_error_handler
{
protected:
Internal_error_handler() {}
virtual ~Internal_error_handler() {}
public:
/**
Handle an error condition.
This method can be implemented by a subclass to achieve any of the
following:
- mask an error internally, prevent exposing it to the user,
- mask an error and throw another one instead.
When this method returns true, the error condition is considered
'handled', and will not be propagated to upper layers.
It is the responsability of the code installing an internal handler
to then check for trapped conditions, and implement logic to recover
from the anticipated conditions trapped during runtime.
This mechanism is similar to C++ try/throw/catch:
- 'try' correspond to <code>THD::push_internal_handler()</code>,
- 'throw' correspond to <code>my_error()</code>,
which invokes <code>my_message_sql()</code>,
- 'catch' correspond to checking how/if an internal handler was invoked,
before removing it from the exception stack with
<code>THD::pop_internal_handler()</code>.
@param sql_errno the error number
@param level the error level
@param thd the calling thread
@return true if the error is handled
*/
virtual bool handle_error(uint sql_errno,
MYSQL_ERROR::enum_warning_level level,
THD *thd) = 0;
};
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
@class THD
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
For each client connection we create a separate thread with THD serving as
a thread/connection descriptor
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
A fix and a test case for Bug#10760 and complementary cleanups. The idea of the patch is that every cursor gets its own lock id for table level locking. Thus cursors are protected from updates performed within the same connection. Additionally a list of transient (must be closed at commit) cursors is maintained and all transient cursors are closed when necessary. Lastly, this patch adds support for deadlock timeouts to TLL locking when using cursors. + post-review fixes.
2005-07-19 20:21:12 +02:00
class THD :public Statement,
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
public Open_tables_state
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
{
Import changeset
2000-07-31 21:29:14 +02:00
public:
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
/* Used to execute base64 coded binlog events in MySQL server */
RELAY_LOG_INFO* rli_fake;
Fix for BUG#13037: undefined variable in IF cause erroneous error-message.
2005-10-25 11:02:48 +02:00
/*
Constant for THD::where initialization in the beginning of every query.
It's needed because we do not save/restore THD::where normally during
primary (non subselect) query execution.
*/
static const char * const DEFAULT_WHERE;
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
#ifdef EMBEDDED_LIBRARY
struct st_mysql *mysql;
SCRUM Prepared statements in embedded server Several changes in library code with two goals: to make mysql_prepare_stmt working in embedded server to get rid of #define mysql_interface_func mysql->methods->interface_func in user's interface
2003-09-16 13:06:25 +02:00
unsigned long client_stmt_id;
unsigned long client_param_count;
SCRUM: prepared statements in embedded library
2003-09-17 17:48:53 +02:00
struct st_mysql_bind *client_params;
Fix for #1487
2003-10-06 13:32:38 +02:00
char *extra_data;
ulong extra_length;
Big patch to make embedded-server working in 5.x Now it supports queries returning several results (particularly important with the SP)
2006-01-04 11:20:28 +01:00
struct st_mysql_data *cur_data;
struct st_mysql_data *first_data;
struct st_mysql_data **data_tail;
void clear_data_list();
struct st_mysql_data *alloc_new_dataset();
Bug #23427 (incompatible ABI change) the incompatibility was caused by current_stmt member added to the MYSQL structure. It's possible to move it to THD structure instead which saves ABI
2006-10-24 14:19:02 +02:00
/*
In embedded server it points to the statement that is processed
in the current query. We store some results directly in statement
fields then.
*/
struct st_mysql_stmt *current_stmt;
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
#endif
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
NET net; // client connection descriptor
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
MEM_ROOT warn_root; // For warnings and errors
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
Protocol *protocol; // Current protocol
Mechanical class renaming: Protocol_simple->Protocol_text; Protocol_prep->Protocol_binary and also THD::protocol_simple->THD::protocol_text, THD::protocol_prep->THD::protocol_binary. Reason: the binary protocol is not bound to be used only with prepared statements long term (see WL#3559 "Decouple binary protocol from prepared statements"). Renaming now is pressing because the fix for BUG#735 "Prepared Statements: there is no support for Query Cache" will introduce a new member in class Query_cache_flags telling about the protocol's nature. Other reason: "simple" is less accurate than "text". Future patches for BUG#735 will rely on this cset.
2007-01-30 22:48:05 +01:00
Protocol_text protocol_text; // Normal protocol
Protocol_binary protocol_binary; // Binary protocol
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
HASH user_vars; // hash for user variables
String packet; // dynamic buffer for network I/O
Support for character set conversion in binary protocol: another go after Monty's review. - Item_param was rewritten. - it turns out that we can't convert string data to character set of connection on the fly, because they first should be written to the binary log. To support efficient conversion we need to rewrite prepared statements binlogging code first.
2004-05-25 00:03:49 +02:00
String convert_buffer; // buffer for charset conversions
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
struct sockaddr_in remote; // client socket address
struct rand_struct rand; // used for authentication
struct system_variables variables; // Changeable local variables
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
struct system_status_var status_var; // Per thread statistic vars
SHOW STATUS does not anymore change local status variables (except com_show_status). Global status variables are still updated. SHOW STATUS are not anymore put in slow query log because of no index usage. Implemntation done by removing orig_sql_command and moving logic of SHOW STATUS to mysql_excute_command() This simplifies code and allows us to remove some if statements all over the code. Upgraded uc_update_queries[] to sql_command_flags and added more bitmaps to better categorize commands. This allowed some overall simplifaction when testing sql_command. Fixes bugs: Bug#10210: running SHOW STATUS increments counters it shouldn't Bug#19764: SHOW commands end up in the slow log as table scans
2006-06-20 12:20:32 +02:00
struct system_status_var *initial_status_var; /* used by show status */
A fix and a test case for Bug#10760 and complementary cleanups. The idea of the patch is that every cursor gets its own lock id for table level locking. Thus cursors are protected from updates performed within the same connection. Additionally a list of transient (must be closed at commit) cursors is maintained and all transient cursors are closed when necessary. Lastly, this patch adds support for deadlock timeouts to TLL locking when using cursors. + post-review fixes.
2005-07-19 20:21:12 +02:00
THR_LOCK_INFO lock_info; // Locking info of this thread
THR_LOCK_OWNER main_lock_id; // To use for conventional queries
THR_LOCK_OWNER *lock_id; // If not main_lock_id, points to
// the lock_id of a cursor.
Fixed bug in wait_for_update() that I had introduced. Changed option variables to my_bool (to avoid bugs in my_getopt()) Added new thread specific mutex LOCK_delete to be able to free LOCK_thread_count early. Changed usage of LOCK_thread_count -> LOCK_status for statistics variables
2002-08-22 15:50:58 +02:00
pthread_mutex_t LOCK_delete; // Locked before thd is deleted
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
/* all prepared statements and cursors of this connection */
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
Statement_map stmt_map;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
/*
A pointer to the stack frame of handle_one_connection(),
which is called first in the thread for handling a client
*/
char *thread_stack;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
/*
db - currently selected database
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
catalog - currently selected catalog
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
WARNING: some members of THD (currently 'db', 'catalog' and 'query') are
set and alloced by the slave SQL thread (for the THD of that thread); that
thread is (and must remain, for now) the only responsible for freeing these
3 members. If you add members here, and you add code to set them in
replication, don't forget to free_them_and_set_them_to_0 in replication
pthread_handler_decl() changed to be ctags-friendly (and contain extern "C" when necessary)
2005-10-08 16:39:55 +02:00
properly. For details see the 'err:' label of the handle_slave_sql()
in sql/slave.cc.
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
*/
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
char *db, *catalog;
WL#2787 (Add view definer/owner to the view definition (.frm) to check privileges on used tables and stored routines when using a VIEW.) Part 2 postreview fixes.
2005-09-20 20:20:38 +02:00
Security_context main_security_ctx;
Security_context *security_ctx;
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
/* remote (peer) port */
uint16 peer_port;
comments about designation of thd->where and thd->proc_info added
2003-10-30 19:30:20 +01:00
/*
Points to info-string that we show in SHOW PROCESSLIST
You are supposed to update thd->proc_info only if you have coded
a time-consuming piece that MySQL can get stuck in for a long time.
*/
Added the following new privleges: SHOW DATABASES CREATE TEMPORARY TABLE LOCK TABLES REPLICATION SLAVE & REPLICATION CLIENT SUPER EXECUTE All scripts & documentation is updated for this change. Added better error messages for global privileges
2002-06-12 14:04:18 +02:00
const char *proc_info;
CURRENT_USER() and "access denied" error messages now report hostname exactly as it was specified in the GRANT command (with wildcards, that is)
2003-04-29 00:15:18 +02:00
Change client_flag to unsigned long (16 -> 32 bits) to handle more options. Don't use new password format if mysql.user has old format tables_priv was not reset on FLUSH PRIVILEGES if tables_priv was empty Portability fixes for Windows
2003-01-18 15:39:21 +01:00
ulong client_capabilities; /* What the client supports */
Added support for max_allowed_packet in option files read by mysql_option() Extended max_allowed_packet for clients to 1G Fixed bug in sending compressed rows >= 16M Fix bug in skiping too long packets from clients. Added checking of wrong command number sent by client.
2003-01-04 14:17:16 +01:00
ulong max_client_packet_length;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Merge with 4.0 for 4.1 release Noteworthy: - New HANDLER code - New multi-update-grant-check code - Table lock code in ha_innodb.cc was not applied
2004-10-06 18:14:33 +02:00
HASH handler_tables_hash;
comment to user-level lock added
2003-12-04 17:12:01 +01:00
/*
One thread can hold up to one named user-level lock. This variable
points to a lock object if the lock is present. See item_func.cc and
chapter 'Miscellaneous functions', for functions GET_LOCK, RELEASE_LOCK.
*/
Fix of compilation failure of latest 4.1 tree: new constants in mysqld.cc exceed unsigned long limit.
2004-03-15 20:39:36 +01:00
User_level_lock *ull;
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
#ifndef DBUG_OFF
uint dbug_sentry; // watch out for memory corruption
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
#endif
Import changeset
2000-07-31 21:29:14 +02:00
struct st_my_thread_var *mysys_var;
Desperate attempt to push part of prepared statements cleanup which was reviewed in Saint-Petersbourg (including post-review fixes).
2004-03-02 20:39:50 +01:00
/*
Rename all prepared statements COM_ commands to prefix with COM_STMT_
2005-06-17 21:26:25 +02:00
Type of current query: COM_STMT_PREPARE, COM_QUERY, etc. Set from
Desperate attempt to push part of prepared statements cleanup which was reviewed in Saint-Petersbourg (including post-review fixes).
2004-03-02 20:39:50 +01:00
first byte of the packet in do_command()
*/
enum enum_server_command command;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
uint32 server_id;
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
uint32 file_id; // for LOAD DATA INFILE
comments about designation of thd->where and thd->proc_info added
2003-10-30 19:30:20 +01:00
/*
Used in error messages to tell user in what part of MySQL we found an
error. E. g. when where= "having clause", if fix_fields() fails, user
will know that the error was in having clause.
*/
Import changeset
2000-07-31 21:29:14 +02:00
const char *where;
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
time_t start_time,time_after_lock,user_time;
time_t connect_time,thr_create_time; // track down slow pthread_create
Import changeset
2000-07-31 21:29:14 +02:00
thr_lock_type update_lock_default;
Follow the coding style with class names.
2007-05-10 16:27:36 +02:00
Delayed_insert *di;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
/* <> 0 if we are inside of trigger or stored function. */
uint in_sub_stmt;
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
/* container for handler's private per-connection data */
void *ha_data[MAX_HA];
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
#ifndef MYSQL_CLIENT
Bug#16837 (Missing #ifdef cause compile problem --without-row-based-replication): Added #ifdef's to make code work even when the system is built without row-based replication.
2006-05-16 11:16:23 +02:00
int binlog_setup_trx_data();
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
/*
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
Public interface to write RBR events to the binlog
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
*/
I had forgotten to delete an already disabled line of C++ code. --- BUG#20265 (Replication of CREATE-SELECT does not work correctly): Fixing bug by making binary log handle statement transactions. The binary log transaction cache can now be truncated to remove events inserted during this statement or transaction. Also, the binary log participate in XA transaction handling, although not as a full 2pc resource.
2006-10-05 10:46:14 +02:00
void binlog_start_trans_and_stmt();
BUG#22864 (Rollback following CREATE... SELECT discards 'CREATE TABLE' from log): When row-based logging is used, the CREATE-SELECT is written as two parts: as a CREATE TABLE statement and as the rows for the table. For both transactional and non-transactional tables, the CREATE TABLE statement was written to the transaction cache, as were the rows, and on statement end, the entire transaction cache was written to the binary log if the table was non-transactional. For transactional tables, the events were kept in the transaction cache until end of transaction (or statement that were not part of a transaction). For the case when AUTOCOMMIT=0 and we are creating a transactional table using a create select, we would then keep the CREATE TABLE statement and the rows for the CREATE-SELECT, while executing the following statements. On a rollback, the transaction cache would then be cleared, which would also remove the CREATE TABLE statement. Hence no table would be created on the slave, while there is an empty table on the master. This relates to BUG#22865 where the table being created exists on the master, but not on the slave during insertion of rows into the newly created table. This occurs since the CREATE TABLE statement were still in the transaction cache until the statement finished executing, and possibly longer if the table was transactional. This patch changes the behaviour of the CREATE-SELECT statement by adding an implicit commit at the end of the statement when creating non-temporary tables. Hence, non-temporary tables will be written to the binary log on completion, and in the even of AUTOCOMMIT=0, a new transaction will be started. Temporary tables do not commit an ongoing transaction: neither as a pre- not a post-commit. The events for both transactional and non-transactional tables are saved in the transaction cache, and written to the binary log at end of the statement.
2006-12-21 09:29:02 +01:00
int binlog_flush_transaction_cache();
void binlog_set_stmt_begin();
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
int binlog_write_table_map(TABLE *table, bool is_transactional);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
int binlog_write_row(TABLE* table, bool is_transactional,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
MY_BITMAP const* cols, size_t colcnt,
const uchar *buf);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
int binlog_delete_row(TABLE* table, bool is_transactional,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
MY_BITMAP const* cols, size_t colcnt,
const uchar *buf);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
int binlog_update_row(TABLE* table, bool is_transactional,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
MY_BITMAP const* cols, size_t colcnt,
const uchar *old_data, const uchar *new_data);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
void set_server_id(uint32 sid) { server_id = sid; }
/*
Member functions to handle pending event for row-level logging.
*/
template <class RowsEventT> Rows_log_event*
binlog_prepare_pending_rows_event(TABLE* table, uint32 serv_id,
MY_BITMAP const* cols,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
size_t colcnt,
size_t needed,
Changes to support aCC on HP-UX.
2006-01-09 15:59:39 +01:00
bool is_transactional,
RowsEventT* hint);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
Rows_log_event* binlog_get_pending_rows_event() const;
void binlog_set_pending_rows_event(Rows_log_event* ev);
int binlog_flush_pending_rows_event(bool stmt_end);
void binlog_delete_pending_rows_event();
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
private:
uint binlog_table_maps; // Number of table maps currently in the binlog
WL#3339 (Issue warnings when statement-based replication may fail): Replacing binlog_row_based_if_mixed with variable binlog_stmt_flags holding several flags and adding member functions to manipulate the flags. Added code to generate a warning when an attempt to log an unsafe statement to the binary log was made. The warning is both pushed to the SHOW WARNINGS table and written to the error log. The prevent flooding the error log, the warning is just written to the error log once per open session.
2007-05-14 14:45:38 +02:00
enum enum_binlog_flag {
BINLOG_FLAG_UNSAFE_STMT_PRINTED,
BINLOG_FLAG_COUNT
};
/**
Flags with per-thread information regarding the status of the
binary log.
*/
uint32 binlog_flags;
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
public:
Bug#19995 (Extreneous table maps generated for statements that do not generate rows): Switched to writing out table maps for tables that are locked when the first row in a statement is seen.
2006-05-31 19:21:52 +02:00
uint get_binlog_table_maps() const {
return binlog_table_maps;
}
Bug#16837 (Missing #ifdef cause compile problem --without-row-based-replication): Added #ifdef's to make code work even when the system is built without row-based replication.
2006-05-16 11:16:23 +02:00
#endif /* MYSQL_CLIENT */
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
public:
Import changeset
2000-07-31 21:29:14 +02:00
struct st_transactions {
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
SAVEPOINT *savepoints;
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
THD_TRANS all; // Trans since BEGIN WORK
THD_TRANS stmt; // Trans for current statement
bug#8151 - truncate leaves a transaction open deadlock in MYSQL_LOG::new_file() style fixes
2005-03-02 10:38:25 +01:00
bool on; // see ha_enable_transaction()
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
XID xid; // transaction identifier
enum xa_states xa_state; // used by external XA only
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
XID_STATE xid_state;
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
Rows_log_event *m_pending_rows_event;
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
/*
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
Tables changed in transaction (that must be invalidated in query cache).
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
List contain only transactional tables, that not invalidated in query
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
cache (instead of full list of changed in transaction tables).
*/
CHANGED_TABLE_LIST* changed_tables;
MEM_ROOT mem_root; // Transaction-life memory allocation pool
void cleanup()
{
forget savepoints at the end of transaction
2005-02-22 21:57:57 +01:00
changed_tables= 0;
savepoints= 0;
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
#ifdef USING_TRANSACTIONS
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
free_root(&mem_root,MYF(MY_KEEP_PREALLOC));
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
#endif
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
}
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
st_transactions()
{
Fix for bug #15047: "server crash when compiling without transaction support".
2005-11-21 08:52:58 +01:00
#ifdef USING_TRANSACTIONS
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
bzero((char*)this, sizeof(*this));
BUG#12162 - one can start two transactions with the same XID. Now we keep all active XID's in a hash
2005-08-12 21:15:01 +02:00
xid_state.xid.null();
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
init_sql_alloc(&mem_root, ALLOC_ROOT_MIN_BLOCK_SIZE, 0);
Fix for bug #15047: "server crash when compiling without transaction support".
2005-11-21 08:52:58 +01:00
#else
xid_state.xa_state= XA_NOTR;
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
#endif
Fix for bug #15047: "server crash when compiling without transaction support".
2005-11-21 08:52:58 +01:00
}
Import changeset
2000-07-31 21:29:14 +02:00
} transaction;
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
Field *dup_field;
Import changeset
2000-07-31 21:29:14 +02:00
#ifndef __WIN__
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
sigset_t signals;
Import changeset
2000-07-31 21:29:14 +02:00
#endif
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
#ifdef SIGNAL_WITH_VIO_CLOSE
Vio* active_vio;
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
#endif
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
/*
This is to track items changed during execution of a prepared
statement/stored procedure. It's created by
register_item_tree_change() in memory root of THD, and freed in
Improve a comment.
2005-07-15 11:21:08 +02:00
rollback_item_tree_changes(). For conventional execution it's always
empty.
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
*/
Item_change_list change_list;
fixed subquery with PS (BUG#2462) fixed UNION preparation
2004-02-08 19:14:13 +01:00
/*
Improve a comment.
2005-07-15 11:21:08 +02:00
A permanent memory area of the statement. For conventional
execution, the parsed tree and execution runtime reside in the same
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
memory root. In this case stmt_arena points to THD. In case of
Improve a comment.
2005-07-15 11:21:08 +02:00
a prepared statement or a stored procedure statement, thd->mem_root
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
conventionally points to runtime memory, and thd->stmt_arena
Improve a comment.
2005-07-15 11:21:08 +02:00
points to the memory of the PS/SP, where the parsed tree of the
statement resides. Whenever you need to perform a permanent
transformation of a parsed tree, you should allocate new memory in
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
stmt_arena, to allow correct re-execution of PS/SP.
Note: in the parser, stmt_arena == thd, even for PS/SP.
fixed subquery with PS (BUG#2462) fixed UNION preparation
2004-02-08 19:14:13 +01:00
*/
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
Query_arena *stmt_arena;
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
/* Tells if LAST_INSERT_ID(#) was called for the current statement */
bool arg_of_last_insert_id_function;
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
/*
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
ALL OVER THIS FILE, "insert_id" means "*automatically generated* value for
insertion into an auto_increment column".
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
*/
/*
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
This is the first autogenerated insert id which was *successfully*
inserted by the previous statement (exactly, if the previous statement
didn't successfully insert an autogenerated insert id, then it's the one
of the statement before, etc).
It can also be set by SET LAST_INSERT_ID=# or SELECT LAST_INSERT_ID(#).
It is returned by LAST_INSERT_ID().
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
*/
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
ulonglong first_successful_insert_id_in_prev_stmt;
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
/*
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
Variant of the above, used for storing in statement-based binlog. The
difference is that the one above can change as the execution of a stored
function progresses, while the one below is set once and then does not
change (which is the value which statement-based binlog needs).
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
*/
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
ulonglong first_successful_insert_id_in_prev_stmt_for_binlog;
/*
This is the first autogenerated insert id which was *successfully*
inserted by the current statement. It is maintained only to set
first_successful_insert_id_in_prev_stmt when statement ends.
*/
ulonglong first_successful_insert_id_in_cur_stmt;
/*
We follow this logic:
- when stmt starts, first_successful_insert_id_in_prev_stmt contains the
first insert id successfully inserted by the previous stmt.
- as stmt makes progress, handler::insert_id_for_cur_row changes; every
time get_auto_increment() is called, auto_inc_intervals_for_binlog is
augmented with the reserved interval (if statement-based binlogging).
- at first successful insertion of an autogenerated value,
first_successful_insert_id_in_cur_stmt is set to
handler::insert_id_for_cur_row.
- when stmt goes to binlog, auto_inc_intervals_for_binlog is
binlogged if non-empty.
- when stmt ends, first_successful_insert_id_in_prev_stmt is set to
first_successful_insert_id_in_cur_stmt.
*/
/*
stmt_depends_on_first_successful_insert_id_in_prev_stmt is set when
LAST_INSERT_ID() is used by a statement.
If it is set, first_successful_insert_id_in_prev_stmt_for_binlog will be
stored in the statement-based binlog.
This variable is CUMULATIVE along the execution of a stored function or
trigger: if one substatement sets it to 1 it will stay 1 until the
function/trigger ends, thus making sure that
first_successful_insert_id_in_prev_stmt_for_binlog does not change anymore
and is propagated to the caller for binlogging.
*/
bool stmt_depends_on_first_successful_insert_id_in_prev_stmt;
/*
List of auto_increment intervals reserved by the thread so far, for
storage in the statement-based binlog.
Note that its minimum is not first_successful_insert_id_in_cur_stmt:
assuming a table with an autoinc column, and this happens:
INSERT INTO ... VALUES(3);
SET INSERT_ID=3; INSERT IGNORE ... VALUES (NULL);
then the latter INSERT will insert no rows
(first_successful_insert_id_in_cur_stmt == 0), but storing "INSERT_ID=3"
in the binlog is still needed; the list's minimum will contain 3.
*/
Discrete_intervals_list auto_inc_intervals_in_cur_stmt_for_binlog;
/* Used by replication and SET INSERT_ID */
Discrete_intervals_list auto_inc_intervals_forced;
/*
There is BUG#19630 where statement-based replication of stored
functions/triggers with two auto_increment columns breaks.
We however ensure that it works when there is 0 or 1 auto_increment
column; our rules are
a) on master, while executing a top statement involving substatements,
first top- or sub- statement to generate auto_increment values wins the
fixes after merge. Updates to test's results. We now reset the THD members related to auto_increment+binlog in MYSQL_LOG::write(). This is better than in THD::cleanup_after_query(), which was not able to distinguish between SELECT myfunc1(),myfunc2() and INSERT INTO t SELECT myfunc1(),myfunc2() from a binlogging point of view. Rows_log_event::exec_event() now calls lex_start() instead of mysql_init_query() because the latter now does too much (it resets the binlog format).
2006-07-10 18:41:03 +02:00
exclusive right to see its values be written to binlog (the write
will be done by the statement or its caller), and the losers won't see
their values be written to binlog.
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
b) on slave, while replicating a top statement involving substatements,
first top- or sub- statement to need to read auto_increment values from
the master's binlog wins the exclusive right to read them (so the losers
won't read their values from binlog but instead generate on their own).
a) implies that we mustn't backup/restore
auto_inc_intervals_in_cur_stmt_for_binlog.
b) implies that we mustn't backup/restore auto_inc_intervals_forced.
If there are more than 1 auto_increment columns, then intervals for
different columns may mix into the
auto_inc_intervals_in_cur_stmt_for_binlog list, which is logically wrong,
but there is no point in preventing this mixing by preventing intervals
from the secondly inserted column to come into the list, as such
prevention would be wrong too.
What will happen in the case of
INSERT INTO t1 (auto_inc) VALUES(NULL);
where t1 has a trigger which inserts into an auto_inc column of t2, is
that in binlog we'll store the interval of t1 and the interval of t2 (when
we store intervals, soon), then in slave, t1 will use both intervals, t2
will use none; if t1 inserts the same number of rows as on master,
normally the 2nd interval will not be used by t1, which is fine. t2's
values will be wrong if t2's internal auto_increment counter is different
from what it was on master (which is likely). In 5.1, in mixed binlogging
mode, row-based binlogging is used for such cases where two
auto_increment columns are inserted.
*/
inline void record_first_successful_insert_id_in_cur_stmt(ulonglong id)
{
if (first_successful_insert_id_in_cur_stmt == 0)
first_successful_insert_id_in_cur_stmt= id;
}
inline ulonglong read_first_successful_insert_id_in_prev_stmt(void)
{
if (!stmt_depends_on_first_successful_insert_id_in_prev_stmt)
{
/* It's the first time we read it */
first_successful_insert_id_in_prev_stmt_for_binlog=
first_successful_insert_id_in_prev_stmt;
stmt_depends_on_first_successful_insert_id_in_prev_stmt= 1;
}
return first_successful_insert_id_in_prev_stmt;
}
/*
WL#3464 (Add replication event to denote gap in replication): Fixing automerge problem and updating comments referring to exec_event().
2007-03-30 15:29:30 +02:00
Used by Intvar_log_event::do_apply_event() and by "SET INSERT_ID=#"
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
(mysqlbinlog). We'll soon add a variant which can take many intervals in
argument.
*/
inline void force_one_auto_inc_interval(ulonglong next_id)
{
Post-merge fixes for mysql-5.1-new-rpl
2006-07-12 08:52:47 +02:00
auto_inc_intervals_forced.empty(); // in case of multiple SET INSERT_ID
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
auto_inc_intervals_forced.append(next_id, ULONGLONG_MAX, 0);
}
LAST_INSERT_ID() should not be set if we couldn't generate an auto_increment id.
2003-06-30 12:28:36 +02:00
ulonglong limit_found_rows;
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
ulonglong options; /* Bitmap of states */
longlong row_count_func; /* For the ROW_COUNT() function */
quick fix for found row counter code cleunup
2003-08-20 20:57:37 +02:00
ha_rows cuted_fields,
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
sent_row_count, examined_row_count;
Implementation of WL#2486 - "Process NATURAL and USING joins according to SQL:2003". * Some of the main problems fixed by the patch: - in "select *" queries the * expanded correctly according to ANSI for arbitrary natural/using joins - natural/using joins are correctly transformed into JOIN ... ON for any number/nesting of the joins. - column references are correctly resolved against natural joins of any nesting and combined with arbitrary other joins. * This patch also contains a fix for name resolution of items inside the ON condition of JOIN ... ON - in this case items must be resolved only against the JOIN operands. To support such 'local' name resolution, the patch introduces a stack of name resolution contexts used at parse time. NOTICE: - This patch is not complete in the sense that - there are 2 test cases that still do not pass - one in join.test, one in select.test. Both are marked with a comment "TODO: WL#2486". - it does not include a new test specific for the task
2005-08-12 16:57:19 +02:00
/*
The set of those tables whose fields are referenced in all subqueries
of the query.
TODO: possibly this it is incorrect to have used tables in THD because
with more than one subquery, it is not clear what does the field mean.
*/
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
table_map used_tables;
Features made for Schlund plus several bug fixes. Read a manual for more detail
2002-05-15 12:50:38 +02:00
USER_CONN *user_connect;
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
CHARSET_INFO *db_charset;
after-review fixes
2003-12-04 20:08:26 +01:00
/*
FIXME: this, and some other variables like 'count_cuted_fields'
maybe should be statement/cursor local, that is, moved to Statement
Prepared_statement deployed instead of PREP_STMT.
2003-12-20 00:16:10 +01:00
class. With current implementation warnings produced in each prepared
statement/cursor settle here.
after-review fixes
2003-12-04 20:08:26 +01:00
*/
do not log certain commans in the slow log
2003-11-14 13:50:19 +01:00
List <MYSQL_ERROR> warn_list;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
uint warn_count[(uint) MYSQL_ERROR::WARN_LEVEL_END];
old_total_warn_count is not used and was removed from THD
2003-10-31 13:15:13 +01:00
uint total_warn_count;
Desperate attempt to push part of prepared statements cleanup which was reviewed in Saint-Petersbourg (including post-review fixes).
2004-03-02 20:39:50 +01:00
/*
Id of current query. Statement can be reused to execute several queries
query_id is global in context of the whole MySQL server.
ID is automatically generated from mutex-protected counter.
It's used in handler code for various purposes: to check which columns
from table are necessary for this select, to check if it's necessary to
update auto-updatable fields (like auto_increment and timestamp).
*/
Eliminate warnings noticed by VC7. This includes fixing my_mmap() on Windows to call CreateFileMapping() with correct arguments, and propogating the introduction of query_id_t to everywhere query ids are passed around. (Bug #8826)
2005-03-19 01:12:25 +01:00
query_id_t query_id, warn_id;
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
ulong col_access;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
WL 2826: Seventh step, more fixes for error injects Removed session variables for error injects started using DBUG macros for error injects
2006-02-03 18:05:30 +01:00
#ifdef ERROR_INJECT_SUPPORT
ulong error_inject_value;
#endif
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
/* Statement id is thread-wide. This counter is used to generate ids */
ulong statement_id_counter;
Put temporary files in binlog cache when using BEGIN/COMMIT Let MySQL 4.0 read 4.1 .frm files without 4.1 specific extensions New variables @@rand_seed1 and @@rand_seed2 (used by replication) DROP TEMPORARY TABLE
2002-11-07 03:02:37 +01:00
ulong rand_saved_seed1, rand_saved_seed2;
Support of warnings for all DML statements (Insert, Update and Alter) Fix LOAD DATA INFILE warnings to have a better meanigful messages Fix to make the mysql command line to automatically show the warnings count for all basic commands
2003-04-30 09:02:28 +02:00
ulong row_count; // Row counter, mainly for errors and warnings
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
pthread_t real_id; /* For debugging */
my_thread_id thread_id;
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
uint tmp_table, global_read_lock;
fix for bug #17619 Scheduler race conditions - Scheduler is either initialized at server start or never. Starting & stopping is now suspending & resuming. - The scheduler has clear OO interface - Now all calls to the scheduler are synchronous - GLOBAL event_scheduler uses thd::sys_var_tmp (see set_var.cc) - External API is encapsulated into class Events - Includes fixes for all comments of Kostja's review of 19.05.2005 Starting to merge into 5.1-release (5.1.10) and push
2006-05-22 20:46:13 +02:00
uint server_status,open_options;
enum enum_thread_type system_thread;
Fix compilation failures on Windows caused by the patch for Bug#17199. Fix a minor issue with Bug#16206 (bdb.test failed if the tree is compiled without blackhole).
2006-06-27 12:56:24 +02:00
uint db_length;
new EXPLAIN fixed bug in mysql-test/create-test-result fixed bug in union-subselect engine
2002-09-26 22:08:22 +02:00
uint select_number; //number of select (used for EXPLAIN)
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
/* variables.transaction_isolation is reset to this after each commit */
enum_tx_isolation session_tx_isolation;
Add warnings for single line inserts. To do this I had to convert count_cuted_fields to an enum (to be able to detect setting a NOT NULL field to NULL)
2003-10-11 22:26:39 +02:00
enum_check_fields count_cuted_fields;
Fix for BUG#12637: Make SPs+user variables replication work: * Allocate thd->user_var_events elements on appropriate mem_root * If several SP statements are binlogged as a single statement, collect all user var accesses they make (grep for StoredRoutinesBinlogging for details)
2005-09-07 17:39:47 +02:00
DYNAMIC_ARRAY user_var_events; /* For user variables replication */
MEM_ROOT *user_var_events_alloc; /* Allocate above array elements here */
First version of new authentification procedure: now authentification is one-stage (instead of two-stage in 4.1) For now following tasks have been done: - PASSWORD() function was rewritten. PASSWORD() now returns SHA1 hash_stage2; for new passwords user.password contains '*'hash_stage2; sql_yacc.yy also fixed; - password.c: new functions were implemented, old rolled back to 4.0 state - server code was rewritten to use new authorization algorithm (check_user(), change user, and other stuff in sql/sql_parse.cc) - client code was rewritten to use new authorization algorithm (mysql_real_connect, myslq_authenticate in sql-common/client.c) - now server barks on 45-byte-length 4.1.0 passwords and refuses 4.1.0-style authentification. Users with 4.1.0 passwords are blocked (sql/sql_acl.cc) - mysqladmin.c was fixed to work correctly with new passwords Tests for 4.0-4.1.1, 4.1.1-4.1.1 (with or without db/password) logons was performed; mysqladmin also was tested. Additional check are nevertheless necessary.
2003-07-01 21:40:59 +02:00
Bug#22725 Replication outages from ER_SERVER_SHUTDOWN (1053) set in replication events The reason for the bug was that replaying of a query on slave could not be possible since its event was recorded with the killed error. Due to the specific of handling INSERT, which per-row-while-loop is unbreakable to killing, the query on transactional table should have not appeared in binlog unless there was a call to a stored routine that got interrupted with killing (and then there must be an error returned out of the loop). The offered solution added the following rule for binlogging of INSERT that accounts the above specifics: For INSERT on transactional-table if the error was not set the only raised flag is harmless and is ignored via masking out on time of creation of binlog event. For both table types the combination of raised error and KILLED flag indicates that there was potentially partial execution on master and consistency is under the question. In that case the code continues to binlog an event with an appropriate killed error. The fix relies on the specified behaviour of stored routine that must propagate the error to the top level query handling if the thd->killed flag was raised in the routine execution. The patch adds an arg with the default killed-status-unset value to Query_log_event::Query_log_event.
2007-05-28 21:20:22 +02:00
enum killed_state
{
NOT_KILLED=0,
KILL_BAD_DATA=1,
KILL_CONNECTION=ER_SERVER_SHUTDOWN,
KILL_QUERY=ER_QUERY_INTERRUPTED,
KILLED_NO_VALUE /* means neither of the states */
};
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
killed_state volatile killed;
First version of new authentification procedure: now authentification is one-stage (instead of two-stage in 4.1) For now following tasks have been done: - PASSWORD() function was rewritten. PASSWORD() now returns SHA1 hash_stage2; for new passwords user.password contains '*'hash_stage2; sql_yacc.yy also fixed; - password.c: new functions were implemented, old rolled back to 4.0 state - server code was rewritten to use new authorization algorithm (check_user(), change user, and other stuff in sql/sql_parse.cc) - client code was rewritten to use new authorization algorithm (mysql_real_connect, myslq_authenticate in sql-common/client.c) - now server barks on 45-byte-length 4.1.0 passwords and refuses 4.1.0-style authentification. Users with 4.1.0 passwords are blocked (sql/sql_acl.cc) - mysqladmin.c was fixed to work correctly with new passwords Tests for 4.0-4.1.1, 4.1.1-4.1.1 (with or without db/password) logons was performed; mysqladmin also was tested. Additional check are nevertheless necessary.
2003-07-01 21:40:59 +02:00
/* scramble - random string sent to client on handshake */
few compile-time bugs fixed
2003-07-18 16:57:21 +02:00
char scramble[SCRAMBLE_LENGTH+1];
First version of new authentification procedure: now authentification is one-stage (instead of two-stage in 4.1) For now following tasks have been done: - PASSWORD() function was rewritten. PASSWORD() now returns SHA1 hash_stage2; for new passwords user.password contains '*'hash_stage2; sql_yacc.yy also fixed; - password.c: new functions were implemented, old rolled back to 4.0 state - server code was rewritten to use new authorization algorithm (check_user(), change user, and other stuff in sql/sql_parse.cc) - client code was rewritten to use new authorization algorithm (mysql_real_connect, myslq_authenticate in sql-common/client.c) - now server barks on 45-byte-length 4.1.0 passwords and refuses 4.1.0-style authentification. Users with 4.1.0 passwords are blocked (sql/sql_acl.cc) - mysqladmin.c was fixed to work correctly with new passwords Tests for 4.0-4.1.1, 4.1.1-4.1.1 (with or without db/password) logons was performed; mysqladmin also was tested. Additional check are nevertheless necessary.
2003-07-01 21:40:59 +02:00
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
bool slave_thread, one_shot_set;
WL#2977 and WL#2712 global and session-level variable to set the binlog format (row/statement), and new binlog format called "mixed" (which is statement-based except if only row-based is correct, in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release): SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default; the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha. It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below). The added tests test the possibility or impossibility to SET, their effects, and the mixed mode, including in prepared statements and in stored procedures and functions. Caveats: a) The mixed mode will not work for stored functions: in mixed mode, a stored function will always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()). b) for the same reason, changing the thread's binlog format inside a stored function is refused with an error message. c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask Dmitri). Additionally, as the binlog format is now changeable by each user for his session, I remove the implication which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1 (not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled phantom protection). Plus fixes for compiler warnings.
2006-02-25 22:21:03 +01:00
/* tells if current statement should binlog row-based(1) or stmt-based(0) */
bool current_stmt_binlog_row_based;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
bool locked, some_tables_deleted;
Support of warnings for all DML statements (Insert, Update and Alter) Fix LOAD DATA INFILE warnings to have a better meanigful messages Fix to make the mysql command line to automatically show the warnings count for all basic commands
2003-04-30 09:02:28 +02:00
bool last_cuted_field;
Second attempt: trying to add Statement context to sources. Added classes Statement, Statement_map Merge commit
2003-11-27 18:51:53 +01:00
bool no_errors, password, is_fatal_error;
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
bool query_start_used, rand_used, time_zone_used;
BUG#14553: NULL in WHERE resets LAST_INSERT_ID To make MySQL compatible with some ODBC applications, you can find the AUTO_INCREMENT value for the last inserted row with the following query: SELECT * FROM tbl_name WHERE auto_col IS NULL. This is done with a special code that replaces 'auto_col IS NULL' with 'auto_col = LAST_INSERT_ID'. However this also resets the LAST_INSERT_ID to 0 as it uses it for a flag so as to ensure that only the first SELECT ... WHERE auto_col IS NULL after an INSERT has this special behaviour. In order to avoid resetting the LAST_INSERT_ID a special flag is introduced in the THD class. This flag is used to restrict the second and subsequent SELECTs instead of LAST_INSERT_ID.
2006-07-10 15:27:03 +02:00
/* for IS NULL => = last_insert_id() fix in remove_eq_conds() */
bool substitute_null_with_insert_id;
Making FLUSH TABLES WITH READ LOCK block COMMITs of existing transactions, in a deadlock-free manner. This splits locking the global read lock in two steps. This fixes a consequence of this bug, known as: BUG#4953 'mysqldump --master-data may report incorrect binlog position if using InnoDB' And a test.
2004-08-20 16:35:23 +02:00
bool in_lock_tables;
Portability fixes. Patches required by Gemini Fix to properly detect if there is an active transaction in InnoDB Fix to not lock thread structure when doing automatic rollback when thread ends Allow -O lower_case_names=0 on UNIX
2001-08-21 19:06:00 +02:00
bool query_error, bootstrap, cleanup_done;
Move tmp_table_used to THD Optimize depending sub querys Remove valgrind warnings
2003-02-10 16:59:16 +01:00
bool tmp_table_used;
After merge fixes Use server character set if --default-character-set is not used Added convert_string() for more efficient alloc+character-set convert of strings
2003-08-18 23:08:08 +02:00
bool charset_is_system_charset, charset_is_collation_connection;
loaddata.result, loaddata.test: Adding test case. sql_yacc.yy: Adding TEXT_STRING_filesystem, which converts from character_set_client to character_set_conversion. Replacing TEXT_STRING_sys to TEXT_STRING_filesystem in LOAD DATA and SELECT INTO OUTFILE contexts. sql_class.h, sql_class.cc: Adding character_set_filesystem variable, and charset_is_character_set_filesystem flag (to avoid conversion when it's not necessary). set_var.h, set_var.cc: Adding sys_var_character_set_filesystem mysqld.cc: Adding --character-set-filesystem startup option.
2006-01-18 09:55:38 +01:00
bool charset_is_character_set_filesystem;
A fix for Bug#9141 "4.1 does not log into slow log ALTER, OPTIMIZE and ANALYZE statements". In 4.1 we disabled logging of slow admin statements. The fix adds an option to enable it back. No test case (slow log is not tested in the test suite), but tested manually. + post-review fixes (word police mainly).
2005-06-16 21:05:38 +02:00
bool enable_slow_log; /* enable slow log for current statement */
Bug #27395 OPTION_STATUS_NO_TRANS_UPDATE is not preserved at the end of SF() thd->options' OPTION_STATUS_NO_TRANS_UPDATE bit was not restored at the end of SF() invocation, where SF() modified non-ta table. As the result of this artifact it was not possible to detect whether there were any side-effects when top-level query ends. If the top level query table was not modified and the bit is lost there would be no binlogging. Fixed with preserving the bit inside of thd->no_trans_update struct. The struct agregates two bool flags telling whether the current query and the current transaction modified any non-ta table. The flags stmt, all are dropped at the end of the query and the transaction.
2007-03-23 16:12:58 +01:00
struct {
bool all:1;
bool stmt:1;
} no_trans_update;
bool abort_on_warning;
Merge
2005-02-11 22:33:52 +01:00
bool got_warning; /* Set on call to push_warning() */
Show all generated warnings in SHOW ERROR Previously we only stored the first given error (the error sent to the client)
2005-02-24 22:33:42 +01:00
bool no_warnings_for_error; /* no warnings on call to my_error() */
fixed mechanism of detection selection from table wich we update (BUG##9398, BUG#8703) fixed wrong join view detection in multi-delete which lead to server crash
2005-03-28 14:13:31 +02:00
/* set during loop of derived table processing */
bool derived_tables_processing;
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
my_bool tablespace_op; /* This is TRUE in DISCARD/IMPORT TABLESPACE */
Simplistic, experimental framework for Stored Procedures (SPs). Implements creation and dropping of PROCEDUREs, IN, OUT, and INOUT parameters, single-statement procedures, rudimentary multi-statement (begin-end) prodedures (when the client can handle it), and local variables. Missing most of the embedded SQL language, all attributes, FUNCTIONs, error handling, reparses procedures at each call (no caching), etc, etc. Certainly buggy too, but procedures can actually be created and called....
2002-12-08 19:59:22 +01:00
sp_rcontext *spcont; // SP runtime context
Code cleanup (and moved sp cache to separate file).
2003-07-03 15:58:37 +02:00
sp_cache *sp_proc_cache;
sp_cache *sp_func_cache;
sql_error.cc, sql_prepare.cc: new file Client-server protocol 4.1 changes - Server side: * Enhanced metadata information: - SHOW [COUNT(*)] ERRORS [LIMIT [offset,] rows] - SHOW [COUNT(*)] WARNING [LIMIT [offset,] rows] - SHOW TABLE TYPES - SHOW PRIVILEGES - SHOW COLUMN TYPES (Not fully implemented) * Prepared execution * Long data handling in pieces * And other misc changes
2002-06-12 23:13:12 +02:00
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
/*
If we do a purge of binary logs, log index info of the threads
that are currently reading it needs to be adjusted. To do that
each thread that is using LOG_INFO needs to adjust the pointer to it
*/
fixed --skip-slave-thread bug added PURGE MASTER LOGS TO and SHOW MASTER LOGS fixed the output of SHOW MASTER STATUS updated docs
2000-10-27 06:11:55 +02:00
LOG_INFO* current_linfo;
Update of query cache code. Changed some sql_alloc() -> thd->alloc() Removed a lot of compiler warnings on Linux Alpha (64 bit) Fixed some core dumps on 64 bit systems (wrong type for packet_len)
2001-12-05 12:03:00 +01:00
NET* slave_net; // network connection from slave -> m.
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
/* Used by the sys_var class to store temporary values */
union
{
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
my_bool my_bool_value;
long long_value;
ulong ulong_value;
ulonglong ulonglong_value;
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
} sys_var_tmp;
Fix for BUG#12335 (SP replication) : New binlogging strategy for stored PROCEDUREs/FUNCTIONs. "Interleaved SPs execution is now binlogged properly, "SELECT spfunc()" is binlogged too. The known remaining issue is binlogging/replication of "a routine is deleted while it is executed" scenario.
2005-08-25 15:34:34 +02:00
struct {
/*
If true, mysql_bin_log::write(Log_event) call will not write events to
binlog, and maintain 2 below variables instead (use
mysql_bin_log.start_union_events to turn this on)
*/
bool do_union;
/*
If TRUE, at least one mysql_bin_log::write(Log_event) call has been
made after last mysql_bin_log.start_union_events() call.
*/
bool unioned_events;
/*
If TRUE, at least one mysql_bin_log::write(Log_event e), where
e.cache_stmt == TRUE call has been made after last
mysql_bin_log.start_union_events() call.
*/
bool unioned_events_trans;
Fix for BUG#12637: Make SPs+user variables replication work: * Allocate thd->user_var_events elements on appropriate mem_root * If several SP statements are binlogged as a single statement, collect all user var accesses they make (grep for StoredRoutinesBinlogging for details)
2005-09-07 17:39:47 +02:00
/*
'queries' (actually SP statements) that run under inside this binlog
union have thd->query_id >= first_query_id.
*/
query_id_t first_query_id;
Fix for BUG#12335 (SP replication) : New binlogging strategy for stored PROCEDUREs/FUNCTIONs. "Interleaved SPs execution is now binlogged properly, "SELECT spfunc()" is binlogged too. The known remaining issue is binlogging/replication of "a routine is deleted while it is executed" scenario.
2005-08-25 15:34:34 +02:00
} binlog_evt_union;
manual merge 5.0-runtime -> 5.1->runtime, with 25411 part I
2007-04-26 05:38:12 +02:00
/**
Character input stream consumed by the lexical analyser,
used during parsing.
Note that since the parser is not re-entrant, we keep only one input
stream here. This member is valid only when executing code during parsing,
and may point to invalid memory after that.
*/
Lex_input_stream *m_lip;
Code #ifdef-ed
2006-03-20 15:46:13 +01:00
#ifdef WITH_PARTITION_STORAGE_ENGINE
bug 17290 and bug 14350 added THD::work_part_info member where we now store modified partition_info structure. It allows no solve problem when different parts of the part_info get into different mem_roots
2006-03-18 15:48:21 +01:00
partition_info *work_part_info;
Code #ifdef-ed
2006-03-20 15:46:13 +01:00
#endif
foo1
2006-04-13 09:50:33 +02:00
Import changeset
2000-07-31 21:29:14 +02:00
THD();
~THD();
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
Small improvement to alloc_root Add support for LIMIT # OFFSET # Changed lock handling: Now all locks should be stored in TABLE_LIST instead of passed to functions. Don't call query_cache_invalidate() twice in some cases mysql_change_user() now clears states to be equal to close + connect. Fixed a bug with multi-table-update and multi-table-delete when used with LOCK TABLES Fixed a bug with replicate-do and UPDATE
2002-11-16 19:19:10 +01:00
void init(void);
THD::init_for_queries() pushed back: see comments to the method why
2003-12-21 20:26:45 +01:00
/*
Initialize memory roots necessary for query processing and (!)
pre-allocate memory for it. We can't do that in THD constructor because
there are use cases (acl_init, delayed inserts, watcher threads,
killing mysqld) where it's vital to not allocate excessive and not used
memory. Note, that we still don't return error from init_for_queries():
if preallocation fails, we should notice that at the first call to
alloc_root.
*/
void init_for_queries();
Small improvement to alloc_root Add support for LIMIT # OFFSET # Changed lock handling: Now all locks should be stored in TABLE_LIST instead of passed to functions. Don't call query_cache_invalidate() twice in some cases mysql_change_user() now clears states to be equal to close + connect. Fixed a bug with multi-table-update and multi-table-delete when used with LOCK TABLES Fixed a bug with replicate-do and UPDATE
2002-11-16 19:19:10 +01:00
void change_user(void);
Portability fixes. Patches required by Gemini Fix to properly detect if there is an active transaction in InnoDB Fix to not lock thread structure when doing automatic rollback when thread ends Allow -O lower_case_names=0 on UNIX
2001-08-21 19:06:00 +02:00
void cleanup(void);
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
void cleanup_after_query();
Import changeset
2000-07-31 21:29:14 +02:00
bool store_globals();
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
#ifdef SIGNAL_WITH_VIO_CLOSE
inline void set_active_vio(Vio* vio)
untested changes to close socket on slave stop for OS's that do not interrupt I/O after thr_alarm_kill() test case for order by desc coredump
2001-03-13 04:17:32 +01:00
{
Fixed bug in wait_for_update() that I had introduced. Changed option variables to my_bool (to avoid bugs in my_getopt()) Added new thread specific mutex LOCK_delete to be able to free LOCK_thread_count early. Changed usage of LOCK_thread_count -> LOCK_status for statistics variables
2002-08-22 15:50:58 +02:00
pthread_mutex_lock(&LOCK_delete);
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
active_vio = vio;
Fixed bug in wait_for_update() that I had introduced. Changed option variables to my_bool (to avoid bugs in my_getopt()) Added new thread specific mutex LOCK_delete to be able to free LOCK_thread_count early. Changed usage of LOCK_thread_count -> LOCK_status for statistics variables
2002-08-22 15:50:58 +02:00
pthread_mutex_unlock(&LOCK_delete);
untested changes to close socket on slave stop for OS's that do not interrupt I/O after thr_alarm_kill() test case for order by desc coredump
2001-03-13 04:17:32 +01:00
}
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
inline void clear_active_vio()
untested changes to close socket on slave stop for OS's that do not interrupt I/O after thr_alarm_kill() test case for order by desc coredump
2001-03-13 04:17:32 +01:00
{
Fixed bug in wait_for_update() that I had introduced. Changed option variables to my_bool (to avoid bugs in my_getopt()) Added new thread specific mutex LOCK_delete to be able to free LOCK_thread_count early. Changed usage of LOCK_thread_count -> LOCK_status for statistics variables
2002-08-22 15:50:58 +02:00
pthread_mutex_lock(&LOCK_delete);
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
active_vio = 0;
Fixed bug in wait_for_update() that I had introduced. Changed option variables to my_bool (to avoid bugs in my_getopt()) Added new thread specific mutex LOCK_delete to be able to free LOCK_thread_count early. Changed usage of LOCK_thread_count -> LOCK_status for statistics variables
2002-08-22 15:50:58 +02:00
pthread_mutex_unlock(&LOCK_delete);
untested changes to close socket on slave stop for OS's that do not interrupt I/O after thr_alarm_kill() test case for order by desc coredump
2001-03-13 04:17:32 +01:00
}
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
void close_active_vio();
remove the rest of isam/merge references fix a race condition in TC_LOG_BINLOG::unlog
2005-04-05 13:17:49 +02:00
#endif
SCRUM two KILL commands implementation (version 2)
2003-03-31 10:39:46 +02:00
void awake(THD::killed_state state_to_set);
bug#22725 merge with 5.0
2007-05-29 21:17:09 +02:00
#ifndef MYSQL_CLIENT
enum enum_binlog_query_type {
/*
The query can be logged row-based or statement-based
*/
ROW_QUERY_TYPE,
/*
The query has to be logged statement-based
*/
STMT_QUERY_TYPE,
/*
The query represents a change to a table in the "mysql"
database and is currently mapped to ROW_QUERY_TYPE.
*/
MYSQL_QUERY_TYPE,
QUERY_TYPE_COUNT
};
int binlog_query(enum_binlog_query_type qtype,
char const *query, ulong query_len,
bool is_trans, bool suppress_use,
THD::killed_state killed_err_arg= THD::KILLED_NO_VALUE);
#endif
removing assertion (will be moved to 4.1) for non-debug to compile
2004-07-31 09:49:32 +02:00
/*
For enter_cond() / exit_cond() to work the mutex must be got before
remove the rest of isam/merge references fix a race condition in TC_LOG_BINLOG::unlog
2005-04-05 13:17:49 +02:00
enter_cond(); this mutex is then released by exit_cond().
Usage must be: lock mutex; enter_cond(); your code; exit_cond().
removing assertion (will be moved to 4.1) for non-debug to compile
2004-07-31 09:49:32 +02:00
*/
rpl000016.test sync rpl000001.result BitKeeper file /home/sasha/src/bk/mysql/mysql-test/r/rpl000001.result ignore Added BitKeeper/tmp/bkr3sAHD to the ignore list slave.h MASTER_POS_WAIT lex.h MASTER_POS_WAIT slave.cc MASTER_POS_WAIT, do automagic restart on debugging abort, skip rotate events in slave.cc debug abort count sql_repl.cc announce the log name at the start of the log with a fake rotate event item_create.h MASTER_POS_WAIT item_func.cc MASTER_POS_WAIT item_func.h MASTER_POS_WAIT sql_class.h enter_cond(), exit_cond() helper inliners item_create.cc added MASTER_POS_WAIT mysql-test-run.sh speed improvement fixes rpl000007.test sync rpl000003.test sleep -> sync rpl000004.test sleep -> sync, fixed clean up bug rpl000014.test sync rpl000009.test sync rpl000013.test sync rpl000001.test sleep -> sync rpl000008.test sync rpl000006.test sync on cleanup rpl000011.test sync rpl000012.test sync rpl000005.test sleep -> sync rpl000010.test sync rpl000015.test sync rpl000002.test sleep -> sync rpl000014.result we now know the master log name as soon as we connect mysql.cc added optional agrument to --wait mysqltest.c added save_master_pos and sync_with_master commands
2001-01-17 13:47:33 +01:00
inline const char* enter_cond(pthread_cond_t *cond, pthread_mutex_t* mutex,
const char* msg)
{
const char* old_msg = proc_info;
remove the rest of isam/merge references fix a race condition in TC_LOG_BINLOG::unlog
2005-04-05 13:17:49 +02:00
safe_mutex_assert_owner(mutex);
rpl000016.test sync rpl000001.result BitKeeper file /home/sasha/src/bk/mysql/mysql-test/r/rpl000001.result ignore Added BitKeeper/tmp/bkr3sAHD to the ignore list slave.h MASTER_POS_WAIT lex.h MASTER_POS_WAIT slave.cc MASTER_POS_WAIT, do automagic restart on debugging abort, skip rotate events in slave.cc debug abort count sql_repl.cc announce the log name at the start of the log with a fake rotate event item_create.h MASTER_POS_WAIT item_func.cc MASTER_POS_WAIT item_func.h MASTER_POS_WAIT sql_class.h enter_cond(), exit_cond() helper inliners item_create.cc added MASTER_POS_WAIT mysql-test-run.sh speed improvement fixes rpl000007.test sync rpl000003.test sleep -> sync rpl000004.test sleep -> sync, fixed clean up bug rpl000014.test sync rpl000009.test sync rpl000013.test sync rpl000001.test sleep -> sync rpl000008.test sync rpl000006.test sync on cleanup rpl000011.test sync rpl000012.test sync rpl000005.test sleep -> sync rpl000010.test sync rpl000015.test sync rpl000002.test sleep -> sync rpl000014.result we now know the master log name as soon as we connect mysql.cc added optional agrument to --wait mysqltest.c added save_master_pos and sync_with_master commands
2001-01-17 13:47:33 +01:00
mysys_var->current_mutex = mutex;
mysys_var->current_cond = cond;
proc_info = msg;
return old_msg;
}
inline void exit_cond(const char* old_msg)
{
Fix for: Bug #4810 "deadlock with KILL when the victim was in a wait state" (I included mutex unlock into exit_cond() for future safety) and BUG#4827 "KILL while START SLAVE may lead to replication slave crash"
2004-07-31 22:33:20 +02:00
/*
Putting the mutex unlock in exit_cond() ensures that
mysys_var->current_mutex is always unlocked _before_ mysys_var->mutex is
locked (if that would not be the case, you'll get a deadlock if someone
does a THD::awake() on you).
*/
pthread_mutex_unlock(mysys_var->current_mutex);
rpl000016.test sync rpl000001.result BitKeeper file /home/sasha/src/bk/mysql/mysql-test/r/rpl000001.result ignore Added BitKeeper/tmp/bkr3sAHD to the ignore list slave.h MASTER_POS_WAIT lex.h MASTER_POS_WAIT slave.cc MASTER_POS_WAIT, do automagic restart on debugging abort, skip rotate events in slave.cc debug abort count sql_repl.cc announce the log name at the start of the log with a fake rotate event item_create.h MASTER_POS_WAIT item_func.cc MASTER_POS_WAIT item_func.h MASTER_POS_WAIT sql_class.h enter_cond(), exit_cond() helper inliners item_create.cc added MASTER_POS_WAIT mysql-test-run.sh speed improvement fixes rpl000007.test sync rpl000003.test sleep -> sync rpl000004.test sleep -> sync, fixed clean up bug rpl000014.test sync rpl000009.test sync rpl000013.test sync rpl000001.test sleep -> sync rpl000008.test sync rpl000006.test sync on cleanup rpl000011.test sync rpl000012.test sync rpl000005.test sleep -> sync rpl000010.test sync rpl000015.test sync rpl000002.test sleep -> sync rpl000014.result we now know the master log name as soon as we connect mysql.cc added optional agrument to --wait mysqltest.c added save_master_pos and sync_with_master commands
2001-01-17 13:47:33 +01:00
pthread_mutex_lock(&mysys_var->mutex);
mysys_var->current_mutex = 0;
mysys_var->current_cond = 0;
proc_info = old_msg;
pthread_mutex_unlock(&mysys_var->mutex);
}
Import changeset
2000-07-31 21:29:14 +02:00
inline time_t query_start() { query_start_used=1; return start_time; }
Fixes for IO_CACHE
2000-11-16 19:47:28 +01:00
inline void set_time() { if (user_time) start_time=time_after_lock=user_time; else time_after_lock=time(&start_time); }
Fix of LEFT JOIN optimizer bug, problem with key cache on Windows and a lot of portability issues.
2000-10-14 02:16:35 +02:00
inline void end_time() { time(&start_time); }
Fixes for IO_CACHE
2000-11-16 19:47:28 +01:00
inline void set_time(time_t t) { time_after_lock=start_time=user_time=t; }
Faster log::write() call, memory leak fix, flush master fix.
2000-09-16 03:27:21 +02:00
inline void lock_time() { time(&time_after_lock); }
Added SQL_CALC_FOUND_ROWS and FOUND_ROWS()
2001-04-11 23:54:35 +02:00
inline ulonglong found_rows(void)
{
return limit_found_rows;
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
}
All statements are now run in a sub transaction Give warning if we are using non transactional tables and do ROLLBACK Porting of crash-me and benchmarks to FrontBase
2000-11-24 00:51:18 +01:00
inline bool active_transaction()
{
XA (not completely polished out yet)
2005-01-16 13:16:23 +01:00
#ifdef USING_TRANSACTIONS
return server_status & SERVER_STATUS_IN_TRANS;
show_check.test fixed cleanup problem rpl000014.test fixed bug in testcase rename.test fixed cleanup problem handler.cc added a comment sql_class.h fixed bug in active_transaction() when MySQL was not compiled with transaction support
2001-01-03 05:46:33 +01:00
#else
return 0;
#endif
All statements are now run in a sub transaction Give warning if we are using non transactional tables and do ROLLBACK Porting of crash-me and benchmarks to FrontBase
2000-11-24 00:51:18 +01:00
}
new lock for multiupdate: - open and create derived tables - detect which tables should be locked for write - lock and fill derived tables some unitialized variables fixed
2004-11-05 16:29:47 +01:00
inline bool fill_derived_tables()
{
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
return !stmt_arena->is_stmt_prepare() && !lex->only_view_structure();
new lock for multiupdate: - open and create derived tables - detect which tables should be locked for write - lock and fill derived tables some unitialized variables fixed
2004-11-05 16:29:47 +01:00
}
SHOW STATUS does not anymore change local status variables (except com_show_status). Global status variables are still updated. SHOW STATUS are not anymore put in slow query log because of no index usage. Implemntation done by removing orig_sql_command and moving logic of SHOW STATUS to mysql_excute_command() This simplifies code and allows us to remove some if statements all over the code. Upgraded uc_update_queries[] to sql_command_flags and added more bitmaps to better categorize commands. This allowed some overall simplifaction when testing sql_command. Fixes bugs: Bug#10210: running SHOW STATUS increments counters it shouldn't Bug#19764: SHOW commands end up in the slow log as table scans
2006-06-20 12:20:32 +02:00
inline bool fill_information_schema_tables()
{
return !stmt_arena->is_stmt_prepare();
}
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
inline void* trans_alloc(unsigned int size)
new lock for multiupdate: - open and create derived tables - detect which tables should be locked for write - lock and fill derived tables some unitialized variables fixed
2004-11-05 16:29:47 +01:00
{
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
return alloc_root(&transaction.mem_root,size);
}
memory leacks in PS with subqueries fixed (adddition to fix of BUG#2462)
2004-02-12 02:10:26 +01:00
bool convert_string(LEX_STRING *to, CHARSET_INFO *to_cs,
const char *from, uint from_length,
CHARSET_INFO *from_cs);
Support for character set conversion in binary protocol: another go after Monty's review. - Item_param was rewritten. - it turns out that we can't convert string data to character set of connection on the fly, because they first should be written to the binary log. To support efficient conversion we need to rewrite prepared statements binlogging code first.
2004-05-25 00:03:49 +02:00
bool convert_string(String *s, CHARSET_INFO *from_cs, CHARSET_INFO *to_cs);
processing trunsactional tables in query cache
2002-03-15 22:57:31 +01:00
void add_changed_table(TABLE *table);
new function for innodb windows case insensitive tables name work around
2002-09-19 09:36:19 +02:00
void add_changed_table(const char *key, long key_length);
CHANGED_TABLE_LIST * changed_table_dup(const char *key, long key_length);
new EXPLAIN fixed bug in mysql-test/create-test-result fixed bug in union-subselect engine
2002-09-26 22:08:22 +02:00
int send_explain_fields(select_result *result);
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
#ifndef EMBEDDED_LIBRARY
Fixed error handling bug
2002-11-03 23:56:25 +01:00
inline void clear_error()
{
Fixed a lot of compiler warnings (Mainly in mysqld and instance manager) Fixed some possible fatal wrong arguments to printf() style functions Initialized some not initialized variables Fixed bug in stored procedure and continue handlers (Fixes Bug#22150)
2006-11-01 18:41:09 +01:00
DBUG_ENTER("clear_error");
Fixed error handling bug
2002-11-03 23:56:25 +01:00
net.last_error[0]= 0;
net.last_errno= 0;
net.report_error= 0;
errors without code removed net_printf/send_error calls replaced by my_error family functions -1/1 (sent/unsent) error reporting removed (WL#2133)
2004-10-20 03:04:37 +02:00
query_error= 0;
Fixed a lot of compiler warnings (Mainly in mysqld and instance manager) Fixed some possible fatal wrong arguments to printf() style functions Initialized some not initialized variables Fixed bug in stored procedure and continue handlers (Fixes Bug#22150)
2006-11-01 18:41:09 +01:00
DBUG_VOID_RETURN;
Fixed error handling bug
2002-11-03 23:56:25 +01:00
}
Portability fixes (and a typo after last merge)
2004-06-04 10:02:35 +02:00
inline bool vio_ok() const { return net.vio != 0; }
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
#else
void clear_error();
Proposed fix for bug #3412 (embedded server: prepared statement returns empty recordset where some records should be found)
2004-05-28 12:59:29 +02:00
inline bool vio_ok() const { return true; }
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
#endif
fixed bug in determinating uncacheable queries new fatal_error interface to assign is_fatal_error and ne.report_error commant about Item_row
2003-01-30 21:15:44 +01:00
inline void fatal_error()
{
is_fatal_error= 1;
renamed: Item_buff -> Cached_item Item_arena -> Query_arena TEST_ASSERT -> YYERROR_UNLESS
2005-06-15 19:58:35 +02:00
net.report_error= 1;
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
DBUG_PRINT("error",("Fatal error set"));
fixed bug in determinating uncacheable queries new fatal_error interface to assign is_fatal_error and ne.report_error commant about Item_row
2003-01-30 21:15:44 +01:00
}
Some SHOW VARIABLES have been renamed: collation_client -> character_set_client collation_results -> character_set_results character_set -> character_set_server SET NAMES now doesn't start client->server conversion SET CHARACTER SET now starts both client->server and server->client conversion
2003-05-21 14:44:12 +02:00
inline CHARSET_INFO *charset() { return variables.character_set_client; }
After merge fixes Use server character set if --default-character-set is not used Added convert_string() for more efficient alloc+character-set convert of strings
2003-08-18 23:08:08 +02:00
void update_charset();
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
inline Query_arena *activate_stmt_arena_if_needed(Query_arena *backup)
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
{
/*
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
Use the persistent arena if we are in a prepared statement or a stored
procedure statement and we have not already changed to use this arena.
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
*/
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
if (!stmt_arena->is_conventional() && mem_root != stmt_arena->mem_root)
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
{
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
set_n_backup_active_arena(stmt_arena, backup);
return stmt_arena;
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
}
return 0;
}
A small simplification: perform two actions at once, register a change, and perform it (the new Item changes registry).
2004-10-10 01:10:00 +02:00
void change_item_tree(Item **place, Item *new_value)
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
{
A small simplification: perform two actions at once, register a change, and perform it (the new Item changes registry).
2004-10-10 01:10:00 +02:00
/* TODO: check for OOM condition here */
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
if (!stmt_arena->is_conventional())
Simpler arena swapping code Now thd->mem_root is a pointer to thd->main_mem_root and THR_MALLOC is a pointer to thd->mem_root. This gives us the following benefits: - Allow us to easily detect if arena has already been swapped before (this fixes a bug in setup_conds() where arena was swaped twice in some cases) - Faster swaps of arenas (as we don't have to copy the whole MEM_ROOT) - We don't anymore have to call my_pthread_setspecific_ptr(THR_MALLOC,...) to change where memory is alloced. Now it's enough to set thd->mem_root
2004-11-08 00:13:54 +01:00
nocheck_register_item_tree_change(place, *place, mem_root);
A small simplification: perform two actions at once, register a change, and perform it (the new Item changes registry).
2004-10-10 01:10:00 +02:00
*place= new_value;
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
}
void nocheck_register_item_tree_change(Item **place, Item *old_value,
A small simplification: perform two actions at once, register a change, and perform it (the new Item changes registry).
2004-10-10 01:10:00 +02:00
MEM_ROOT *runtime_memroot);
A fix for Bug#5748 "Prepared statement with BETWEEN and bigint values crashes mysqld": implementation for a generic item tree modifications registry. Every item tree modification which should be rolled back for subsequent execution of a prepared statement or stored procedure should be saved in the registry. All such modifications are rolled back at once during cleanup stage of PS. Actual fix for the bug just adds a call to register modifications to convert_constant_item. Post review fixes implemented.
2004-10-08 00:21:19 +02:00
void rollback_item_tree_changes();
A fix and test case for Bug#5985 ""prepare stmt from "select rand(?)" crashes server." The fix makes Item_func_rand prepared-statements aware plus it fixes the case when RAND is used in prepared statements and replication is on (as well as several similar issues). Until now we did not reset THD before every execution of a prepared statement, so if some execution had set thd->time_zone_used or thd->rand_used they would not be reset until next mysql_parse. Some of post-review fixes done.
2004-10-14 00:53:59 +02:00
/*
Cleanup statement parse state (parse tree, lex) and execution
state after execution of a non-prepared SQL statement.
*/
void end_statement();
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
inline int killed_errno() const
{
Bug#22725 Replication outages from ER_SERVER_SHUTDOWN (1053) set in replication events The reason for the bug was that replaying of a query on slave could not be possible since its event was recorded with the killed error. Due to the specific of handling INSERT, which per-row-while-loop is unbreakable to killing, the query on transactional table should have not appeared in binlog unless there was a call to a stored routine that got interrupted with killing (and then there must be an error returned out of the loop). The offered solution added the following rule for binlogging of INSERT that accounts the above specifics: For INSERT on transactional-table if the error was not set the only raised flag is harmless and is ignored via masking out on time of creation of binlog event. For both table types the combination of raised error and KILLED flag indicates that there was potentially partial execution on master and consistency is under the question. In that case the code continues to binlog an event with an appropriate killed error. The fix relies on the specified behaviour of stored routine that must propagate the error to the top level query handling if the thd->killed flag was raised in the routine execution. The patch adds an arg with the default killed-status-unset value to Query_log_event::Query_log_event.
2007-05-28 21:20:22 +02:00
killed_state killed_val; /* to cache the volatile 'killed' */
return (killed_val= killed) != KILL_BAD_DATA ? killed_val : 0;
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
}
inline void send_kill_message() const
{
post-review fixes
2004-11-12 13:34:00 +01:00
int err= killed_errno();
Don't generate message in send_kill_message() if wrong data. (Caused core dump for wrong data in ALTER TABLE on some system)
2005-04-05 00:19:48 +02:00
if (err)
my_message(err, ER(err), MYF(0));
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
}
Strict mode & better warnings Under strict mode MySQL will generate an error message if there was any conversion when assigning data to a field. Added checking of date/datetime fields. If strict mode, give error if we have not given value to field without a default value (for INSERT)
2004-09-28 19:08:00 +02:00
/* return TRUE if we will abort query if we make a warning now */
inline bool really_abort_on_warning()
{
return (abort_on_warning &&
Bug #27395 OPTION_STATUS_NO_TRANS_UPDATE is not preserved at the end of SF() thd->options' OPTION_STATUS_NO_TRANS_UPDATE bit was not restored at the end of SF() invocation, where SF() modified non-ta table. As the result of this artifact it was not possible to detect whether there were any side-effects when top-level query ends. If the top level query table was not modified and the bit is lost there would be no binlogging. Fixed with preserving the bit inside of thd->no_trans_update struct. The struct agregates two bool flags telling whether the current query and the current transaction modified any non-ta table. The flags stmt, all are dropped at the end of the query and the transaction.
2007-03-23 16:12:58 +01:00
(!no_trans_update.stmt ||
Strict mode & better warnings Under strict mode MySQL will generate an error message if there was any conversion when assigning data to a field. Added checking of date/datetime fields. If strict mode, give error if we have not given value to field without a default value (for INSERT)
2004-09-28 19:08:00 +02:00
(variables.sql_mode & MODE_STRICT_ALL_TABLES)));
}
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
void set_status_var_init();
stop evaluation constant functions in WHERE (BUG#4663) correct value of CURRENT_USER() in SP with "security definer" (BUG#7291)
2005-07-15 23:01:44 +02:00
bool is_context_analysis_only()
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
{ return stmt_arena->is_stmt_prepare() || lex->view_prepare_mode; }
Fix for bug #10055 "Using stored function with information_schema causes empty result set". To enable full access to contents of I_S tables from stored functions or statements that use them, we manipulate with thread's open tables state and ensure that we won't cause deadlock when we open tables by ignoring flushes and name-locks. Building of contents of I_S.TABLES no longer requires locking of tables since we use use handler::info() method with HA_STATUS_AUTO flag instead of handler::update_auto_increment() for obtaining information about auto-increment values. But this also means that handlers have to implement support for HA_STATUS_AUTO flag (particularly InnoDB needs it).
2005-08-08 15:46:06 +02:00
void reset_n_backup_open_tables_state(Open_tables_state *backup);
void restore_backup_open_tables_state(Open_tables_state *backup);
Save and clear run context before executing a stored function or trigger and restore it afterwards. This allows us to use statement replication with functions and triggers The following things are fixed with this patch: - NOW() and automatic timestamps takes the value from the main event for functions and triggers (which allows these to replicate with statement level logging) - No side effects for triggers or functions with auto-increment values(), last_insert_id(), rand() or found_rows() - Triggers can't return result sets Fixes bugs: #12480: NOW() is not constant in a trigger #12481: Using NOW() in a stored function breaks statement based replication #12482: Triggers has side effects with auto_increment values #11587: trigger causes lost connection error
2005-08-15 17:15:12 +02:00
void reset_sub_statement_state(Sub_statement_state *backup, uint new_state);
void restore_sub_statement_state(Sub_statement_state *backup);
Rename: - current_arena to stmt_arena: the thread may have more than one 'current' arenas: one for runtime data, and one for the parsed tree of a statement. Only one of them is active at any moment. - set_item_arena -> set_query_arena, because Item_arena was renamed to Query_arena a while ago - set_n_backup_item_arena -> set_n_backup_active_arena; the active arena is the arena thd->mem_root and thd->free_list are currently pointing at. - restore_backup_item_arena -> restore_active_arena (with the same rationale) - change_arena_if_needed -> activate_stmt_arena_if_needed; this method sets thd->stmt_arena active if it's not done yet.
2005-09-02 15:21:19 +02:00
void set_n_backup_active_arena(Query_arena *set, Query_arena *backup);
void restore_active_arena(Query_arena *set, Query_arena *backup);
WL#3339 (Issue warnings when statement-based replication may fail): Replacing binlog_row_based_if_mixed with variable binlog_stmt_flags holding several flags and adding member functions to manipulate the flags. Added code to generate a warning when an attempt to log an unsafe statement to the binary log was made. The warning is both pushed to the SHOW WARNINGS table and written to the error log. The prevent flooding the error log, the warning is just written to the error log once per open session.
2007-05-14 14:45:38 +02:00
WL#2977 and WL#2712 global and session-level variable to set the binlog format (row/statement), and new binlog format called "mixed" (which is statement-based except if only row-based is correct, in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release): SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default; the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha. It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below). The added tests test the possibility or impossibility to SET, their effects, and the mixed mode, including in prepared statements and in stored procedures and functions. Caveats: a) The mixed mode will not work for stored functions: in mixed mode, a stored function will always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()). b) for the same reason, changing the thread's binlog format inside a stored function is refused with an error message. c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask Dmitri). Additionally, as the binlog format is now changeable by each user for his session, I remove the implication which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1 (not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled phantom protection). Plus fixes for compiler warnings.
2006-02-25 22:21:03 +01:00
inline void set_current_stmt_binlog_row_based_if_mixed()
{
* Mixed replication mode * : 1) Fix for BUG#19630 "stored function inserting into two auto_increment breaks statement-based binlog": a stored function inserting into two such tables may fail to replicate (inserting wrong data in the slave's copy of the second table) if the slave's second table had an internal auto_increment counter different from master's. Because the auto_increment value autogenerated by master for the 2nd table does not go into binlog, only the first does, so the slave lacks information. To fix this, if running in mixed binlogging mode, if the stored function or trigger plans to update two different tables both having auto_increment columns, we switch to row-based for the whole function. We don't have a simple solution for statement-based binlogging mode, there the bug remains and will be documented as a known problem. Re-enabling rpl_switch_stm_row_mixed. 2) Fix for BUG#20630 "Mixed binlogging mode does not work with stored functions, triggers, views", which was a documented limitation (in mixed mode, we didn't detect that a stored function's execution needed row-based binlogging (due to some UUID() call for example); same for triggers, same for views (a view created from a SELECT UUID(), and doing INSERT INTO sometable SELECT theview; would not replicate row-based). This is implemented by, after parsing a routine's body, remembering in sp_head that this routine needs row-based binlogging. Then when this routine is used, the caller is marked to require row-based binlogging too. Same for views: when we parse a view and detect that its SELECT needs row-based binary logging, we mark the calling LEX as such. 3) Fix for BUG#20499 "mixed mode with temporary table breaks binlog": a temporary table containing e.g. UUID has its changes not binlogged, so any query updating a permanent table with data from the temporary table will run wrongly on slave. Solution: in mixed mode we don't switch back from row-based to statement-based when there exists temporary tables. 4) Attempt to test mysqlbinlog on a binlog generated by mysqlbinlog; impossible due to BUG#11312 and BUG#20329, but test is in place for when they are fixed.
2006-07-09 17:00:47 +02:00
/*
If in a stored/function trigger, the caller should already have done the
change. We test in_sub_stmt to prevent introducing bugs where people
wouldn't ensure that, and would switch to row-based mode in the middle
of executing a stored function/trigger (which is too late, see also
reset_current_stmt_binlog_row_based()); this condition will make their
tests fail and so force them to propagate the
lex->binlog_row_based_if_mixed upwards to the caller.
*/
if ((variables.binlog_format == BINLOG_FORMAT_MIXED) &&
(in_sub_stmt == 0))
Bug#16837 (Missing #ifdef cause compile problem --without-row-based-replication): Added #ifdef's to make code work even when the system is built without row-based replication.
2006-05-16 11:16:23 +02:00
current_stmt_binlog_row_based= TRUE;
WL#3171: Dynamic logging format Removed error on seeing table map event when slave is in SBR mode and switching temporarily to RBR mode to replicate the rows.
2006-03-08 06:55:21 +01:00
}
inline void set_current_stmt_binlog_row_based()
{
Bug#16837 (Missing #ifdef cause compile problem --without-row-based-replication): Added #ifdef's to make code work even when the system is built without row-based replication.
2006-05-16 11:16:23 +02:00
current_stmt_binlog_row_based= TRUE;
WL#2977 and WL#2712 global and session-level variable to set the binlog format (row/statement), and new binlog format called "mixed" (which is statement-based except if only row-based is correct, in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release): SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default; the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha. It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below). The added tests test the possibility or impossibility to SET, their effects, and the mixed mode, including in prepared statements and in stored procedures and functions. Caveats: a) The mixed mode will not work for stored functions: in mixed mode, a stored function will always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()). b) for the same reason, changing the thread's binlog format inside a stored function is refused with an error message. c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask Dmitri). Additionally, as the binlog format is now changeable by each user for his session, I remove the implication which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1 (not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled phantom protection). Plus fixes for compiler warnings.
2006-02-25 22:21:03 +01:00
}
Bug#19066 (DELETE FROM inconsistency for NDB): Under row-based replication, DELETE FROM will now always be replicated as individual row deletions, while TRUNCATE TABLE will always be replicated as a statement.
2006-06-01 11:53:27 +02:00
inline void clear_current_stmt_binlog_row_based()
{
Merge mysql.com:/home/bkroot/mysql-5.1-new-rpl into mysql.com:/home/bk/b19066-mysql-5.1-new
2006-06-13 22:09:59 +02:00
current_stmt_binlog_row_based= FALSE;
Fixed problem when compiling libmysqld (without replication) This change allows us to use the stmt_binlog function in the code without ifdefs (We should avoid having ifdefs in the .cc and .c files)
2006-06-23 02:21:12 +02:00
}
WL#2977 and WL#2712 global and session-level variable to set the binlog format (row/statement), and new binlog format called "mixed" (which is statement-based except if only row-based is correct, in this cset it means if UDF or UUID is used; more cases could be added in later 5.1 release): SET GLOBAL|SESSION BINLOG_FORMAT=row|statement|mixed|default; the global default is statement unless cluster is enabled (then it's row) as in 5.1-alpha. It's not possible to use SET on this variable if a session is currently in row-based mode and has open temporary tables (because CREATE TEMPORARY TABLE was not binlogged so temp table is not known on slave), or if NDB is enabled (because NDB does not support such change on-the-fly, though it will later), of if in a stored function (see below). The added tests test the possibility or impossibility to SET, their effects, and the mixed mode, including in prepared statements and in stored procedures and functions. Caveats: a) The mixed mode will not work for stored functions: in mixed mode, a stored function will always be binlogged as one call and in a statement-based way (e.g. INSERT VALUES(myfunc()) or SELECT myfunc()). b) for the same reason, changing the thread's binlog format inside a stored function is refused with an error message. c) the same problems apply to triggers; implementing b) for triggers will be done later (will ask Dmitri). Additionally, as the binlog format is now changeable by each user for his session, I remove the implication which was done at startup, where row-based automatically set log-bin-trust-routine-creators to 1 (not possible anymore as a user can now switch to stmt-based and do nasty things again), and automatically set --innodb-locks-unsafe-for-binlog to 1 (was anyway theoretically incorrect as it disabled phantom protection). Plus fixes for compiler warnings.
2006-02-25 22:21:03 +01:00
inline void reset_current_stmt_binlog_row_based()
{
* Mixed replication mode * : 1) Fix for BUG#19630 "stored function inserting into two auto_increment breaks statement-based binlog": a stored function inserting into two such tables may fail to replicate (inserting wrong data in the slave's copy of the second table) if the slave's second table had an internal auto_increment counter different from master's. Because the auto_increment value autogenerated by master for the 2nd table does not go into binlog, only the first does, so the slave lacks information. To fix this, if running in mixed binlogging mode, if the stored function or trigger plans to update two different tables both having auto_increment columns, we switch to row-based for the whole function. We don't have a simple solution for statement-based binlogging mode, there the bug remains and will be documented as a known problem. Re-enabling rpl_switch_stm_row_mixed. 2) Fix for BUG#20630 "Mixed binlogging mode does not work with stored functions, triggers, views", which was a documented limitation (in mixed mode, we didn't detect that a stored function's execution needed row-based binlogging (due to some UUID() call for example); same for triggers, same for views (a view created from a SELECT UUID(), and doing INSERT INTO sometable SELECT theview; would not replicate row-based). This is implemented by, after parsing a routine's body, remembering in sp_head that this routine needs row-based binlogging. Then when this routine is used, the caller is marked to require row-based binlogging too. Same for views: when we parse a view and detect that its SELECT needs row-based binary logging, we mark the calling LEX as such. 3) Fix for BUG#20499 "mixed mode with temporary table breaks binlog": a temporary table containing e.g. UUID has its changes not binlogged, so any query updating a permanent table with data from the temporary table will run wrongly on slave. Solution: in mixed mode we don't switch back from row-based to statement-based when there exists temporary tables. 4) Attempt to test mysqlbinlog on a binlog generated by mysqlbinlog; impossible due to BUG#11312 and BUG#20329, but test is in place for when they are fixed.
2006-07-09 17:00:47 +02:00
/*
If there are temporary tables, don't reset back to
statement-based. Indeed it could be that:
CREATE TEMPORARY TABLE t SELECT UUID(); # row-based
# and row-based does not store updates to temp tables
# in the binlog.
INSERT INTO u SELECT * FROM t; # stmt-based
and then the INSERT will fail as data inserted into t was not logged.
So we continue with row-based until the temp table is dropped.
If we are in a stored function or trigger, we mustn't reset in the
middle of its execution (as the binary logging way of a stored function
or trigger is decided when it starts executing, depending for example on
the caller (for a stored function: if caller is SELECT or
INSERT/UPDATE/DELETE...).
bug#23110 Some NDB tests assert randomly on mixed mode binlog: Don't reset binlog format for NDB binlog injector thread
2006-11-01 22:35:58 +01:00
Don't reset binlog format for NDB binlog injector thread.
* Mixed replication mode * : 1) Fix for BUG#19630 "stored function inserting into two auto_increment breaks statement-based binlog": a stored function inserting into two such tables may fail to replicate (inserting wrong data in the slave's copy of the second table) if the slave's second table had an internal auto_increment counter different from master's. Because the auto_increment value autogenerated by master for the 2nd table does not go into binlog, only the first does, so the slave lacks information. To fix this, if running in mixed binlogging mode, if the stored function or trigger plans to update two different tables both having auto_increment columns, we switch to row-based for the whole function. We don't have a simple solution for statement-based binlogging mode, there the bug remains and will be documented as a known problem. Re-enabling rpl_switch_stm_row_mixed. 2) Fix for BUG#20630 "Mixed binlogging mode does not work with stored functions, triggers, views", which was a documented limitation (in mixed mode, we didn't detect that a stored function's execution needed row-based binlogging (due to some UUID() call for example); same for triggers, same for views (a view created from a SELECT UUID(), and doing INSERT INTO sometable SELECT theview; would not replicate row-based). This is implemented by, after parsing a routine's body, remembering in sp_head that this routine needs row-based binlogging. Then when this routine is used, the caller is marked to require row-based binlogging too. Same for views: when we parse a view and detect that its SELECT needs row-based binary logging, we mark the calling LEX as such. 3) Fix for BUG#20499 "mixed mode with temporary table breaks binlog": a temporary table containing e.g. UUID has its changes not binlogged, so any query updating a permanent table with data from the temporary table will run wrongly on slave. Solution: in mixed mode we don't switch back from row-based to statement-based when there exists temporary tables. 4) Attempt to test mysqlbinlog on a binlog generated by mysqlbinlog; impossible due to BUG#11312 and BUG#20329, but test is in place for when they are fixed.
2006-07-09 17:00:47 +02:00
*/
bug#23110 Some NDB tests assert randomly on mixed mode binlog: Don't reset binlog format for NDB binlog injector thread
2006-11-01 22:35:58 +01:00
if ((temporary_tables == NULL) && (in_sub_stmt == 0) &&
(system_thread != SYSTEM_THREAD_NDBCLUSTER_BINLOG))
* Mixed replication mode * : 1) Fix for BUG#19630 "stored function inserting into two auto_increment breaks statement-based binlog": a stored function inserting into two such tables may fail to replicate (inserting wrong data in the slave's copy of the second table) if the slave's second table had an internal auto_increment counter different from master's. Because the auto_increment value autogenerated by master for the 2nd table does not go into binlog, only the first does, so the slave lacks information. To fix this, if running in mixed binlogging mode, if the stored function or trigger plans to update two different tables both having auto_increment columns, we switch to row-based for the whole function. We don't have a simple solution for statement-based binlogging mode, there the bug remains and will be documented as a known problem. Re-enabling rpl_switch_stm_row_mixed. 2) Fix for BUG#20630 "Mixed binlogging mode does not work with stored functions, triggers, views", which was a documented limitation (in mixed mode, we didn't detect that a stored function's execution needed row-based binlogging (due to some UUID() call for example); same for triggers, same for views (a view created from a SELECT UUID(), and doing INSERT INTO sometable SELECT theview; would not replicate row-based). This is implemented by, after parsing a routine's body, remembering in sp_head that this routine needs row-based binlogging. Then when this routine is used, the caller is marked to require row-based binlogging too. Same for views: when we parse a view and detect that its SELECT needs row-based binary logging, we mark the calling LEX as such. 3) Fix for BUG#20499 "mixed mode with temporary table breaks binlog": a temporary table containing e.g. UUID has its changes not binlogged, so any query updating a permanent table with data from the temporary table will run wrongly on slave. Solution: in mixed mode we don't switch back from row-based to statement-based when there exists temporary tables. 4) Attempt to test mysqlbinlog on a binlog generated by mysqlbinlog; impossible due to BUG#11312 and BUG#20329, but test is in place for when they are fixed.
2006-07-09 17:00:47 +02:00
{
current_stmt_binlog_row_based=
test(variables.binlog_format == BINLOG_FORMAT_ROW);
}
Post-merge fixes for mysql-5.1-new-rpl
2006-07-12 08:52:47 +02:00
}
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
/*
Initialize the current database from a NULL-terminated string with length
A fix for Bug#19022 "Memory bug when switching db during trigger execution". No test case as the bug is in an existing test case (rpl_trigger.test when it is run under valgrind). The warning was caused by memory corruption in replication slave: thd->db was pointing at a stack address that was previously used by sp_head::execute()::old_db. This happened because mysql_change_db behaved differently in replication slave and did not make a copy of the argument to assign to thd->db. The solution is to always free the old value of thd->db and allocate a new copy, regardless whether we're running in a replication slave or not.
2006-06-28 21:47:45 +02:00
If we run out of memory, we free the current database and return TRUE.
This way the user will notice the error as there will be no current
database selected (in addition to the error message set by malloc).
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
*/
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
bool set_db(const char *new_db, size_t new_db_len)
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
{
A fix for Bug#19022 "Memory bug when switching db during trigger execution". No test case as the bug is in an existing test case (rpl_trigger.test when it is run under valgrind). The warning was caused by memory corruption in replication slave: thd->db was pointing at a stack address that was previously used by sp_head::execute()::old_db. This happened because mysql_change_db behaved differently in replication slave and did not make a copy of the argument to assign to thd->db. The solution is to always free the old value of thd->db and allocate a new copy, regardless whether we're running in a replication slave or not.
2006-06-28 21:47:45 +02:00
/* Do not reallocate memory if current chunk is big enough. */
if (db && new_db && db_length >= new_db_len)
memcpy(db, new_db, new_db_len+1);
else
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
{
Post-merge fixes.
2006-07-13 15:34:49 +02:00
x_free(db);
db= new_db ? my_strndup(new_db, new_db_len, MYF(MY_WME)) : NULL;
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
}
A fix for Bug#19022 "Memory bug when switching db during trigger execution". No test case as the bug is in an existing test case (rpl_trigger.test when it is run under valgrind). The warning was caused by memory corruption in replication slave: thd->db was pointing at a stack address that was previously used by sp_head::execute()::old_db. This happened because mysql_change_db behaved differently in replication slave and did not make a copy of the argument to assign to thd->db. The solution is to always free the old value of thd->db and allocate a new copy, regardless whether we're running in a replication slave or not.
2006-06-28 21:47:45 +02:00
db_length= db ? new_db_len : 0;
return new_db && !db;
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
}
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
void reset_db(char *new_db, size_t new_db_len)
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
{
db= new_db;
db_length= new_db_len;
}
/*
Copy the current database to the argument. Use the current arena to
allocate memory for a deep copy: current database may be freed after
a statement is parsed but before it's executed.
*/
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
bool copy_db_to(char **p_db, size_t *p_db_length)
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
{
if (db == NULL)
{
my_message(ER_NO_DB_ERROR, ER(ER_NO_DB_ERROR), MYF(0));
return TRUE;
}
*p_db= strmake(db, db_length);
Changed several char* to LEX_STRING*.
2006-10-16 18:57:33 +02:00
*p_db_length= db_length;
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
return FALSE;
}
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
thd_scheduler scheduler;
Manual merge
2007-03-06 21:46:33 +01:00
public:
/**
Add an internal error handler to the thread execution context.
@param handler the exception handler to add
*/
void push_internal_handler(Internal_error_handler *handler);
/**
Handle an error condition.
@param sql_errno the error number
@param level the error level
@return true if the error is handled
*/
virtual bool handle_error(uint sql_errno,
MYSQL_ERROR::enum_warning_level level);
/**
Remove the error handler last pushed.
*/
void pop_internal_handler();
private:
/** The current internal error handler for this thread, or NULL. */
Internal_error_handler *m_internal_handler;
A fix for Bug#26750 "valgrind leak in sp_head" (and post-review fixes). The legend: on a replication slave, in case a trigger creation was filtered out because of application of replicate-do-table/ replicate-ignore-table rule, the parsed definition of a trigger was not cleaned up properly. LEX::sphead member was left around and leaked memory. Until the actual implementation of support of replicate-ignore-table rules for triggers by the patch for Bug 24478 it was never the case that "case SQLCOM_CREATE_TRIGGER" was not executed once a trigger was parsed, so the deletion of lex->sphead there worked and the memory did not leak. The fix: The real cause of the bug is that there is no 1 or 2 places where we can clean up the main LEX after parse. And the reason we can not have just one or two places where we clean up the LEX is asymmetric behaviour of MYSQLparse in case of success or error. One of the root causes of this behaviour is the code in Item::Item() constructor. There, a newly created item adds itself to THD::free_list - a single-linked list of Items used in a statement. Yuck. This code is unaware that we may have more than one statement active at a time, and always assumes that the free_list of the current statement is located in THD::free_list. One day we need to be able to explicitly allocate an item in a given Query_arena. Thus, when parsing a definition of a stored procedure, like CREATE PROCEDURE p1() BEGIN SELECT a FROM t1; SELECT b FROM t1; END; we actually need to reset THD::mem_root, THD::free_list and THD::lex to parse the nested procedure statement (SELECT *). The actual reset and restore is implemented in semantic actions attached to sp_proc_stmt grammar rule. The problem is that in case of a parsing error inside a nested statement Bison generated parser would abort immediately, without executing the restore part of the semantic action. This would leave THD in an in-the-middle-of-parsing state. This is why we couldn't have had a single place where we clean up the LEX after MYSQLparse - in case of an error we needed to do a clean up immediately, in case of success a clean up could have been delayed. This left the door open for a memory leak. One of the following possibilities were considered when working on a fix: - patch the replication logic to do the clean up. Rejected as breaks module borders, replication code should not need to know the gory details of clean up procedure after CREATE TRIGGER. - wrap MYSQLparse with a function that would do a clean up. Rejected as ideally we should fix the problem when it happens, not adjust for it outside of the problematic code. - make sure MYSQLparse cleans up after itself by invoking the clean up functionality in the appropriate places before return. Implemented in this patch. - use %destructor rule for sp_proc_stmt to restore THD - cleaner than the prevoius approach, but rejected because needs a careful analysis of the side effects, and this patch is for 5.0, and long term we need to use the next alternative anyway - make sure that sp_proc_stmt doesn't juggle with THD - this is a large work that will affect many modules. Cleanup: move main_lex and main_mem_root from Statement to its only two descendants Prepared_statement and THD. This ensures that when a Statement instance was created for purposes of statement backup, we do not involve LEX constructor/destructor, which is fairly expensive. In order to track that the transformation produces equivalent functionality please check the respective constructors and destructors of Statement, Prepared_statement and THD - these members were used only there. This cleanup is unrelated to the patch.
2007-03-07 10:24:46 +01:00
/**
The lex to hold the parsed tree of conventional (non-prepared) queries.
Whereas for prepared and stored procedure statements we use an own lex
instance for each new query, for conventional statements we reuse
the same lex. (@see mysql_parse for details).
*/
LEX main_lex;
/**
This memory root is used for two purposes:
- for conventional queries, to allocate structures stored in main_lex
during parsing, and allocate runtime data (execution plan, etc.)
during execution.
- for prepared queries, only to allocate runtime data. The parsed
tree itself is reused between executions and thus is stored elsewhere.
*/
MEM_ROOT main_mem_root;
Import changeset
2000-07-31 21:29:14 +02:00
};
Implementation of Monty's idea: Now we can open mysql.proc table for lookup of stored routines definitions even if we already have some tables open and locked. To avoid deadlocks in this case we have to put certain restrictions on locking of mysql.proc table. This allows to use stored routines safely under LOCK TABLES without explicitly mentioning mysql.proc in the list of locked tables. It also fixes bug #11554 "Server crashes on statement indirectly using non-cached function".
2005-07-13 11:48:13 +02:00
Manual fixes to merge of fix for Bug #6682 to 4.1.
2004-12-03 02:44:33 +01:00
#define tmp_disable_binlog(A) \
Fixes during review of new pushed code Removed duplicate usage of TMP_TABLE_FORCE_MYISAM by making 'options' longlong
2005-08-12 12:54:42 +02:00
{ulonglong tmp_disable_binlog__save_options= (A)->options; \
fixed environment restoring in case of error during SP function execution (BUG#9503) #define macro improvement
2005-06-23 17:29:10 +02:00
(A)->options&= ~OPTION_BIN_LOG
Manual fixes to merge of fix for Bug #6682 to 4.1.
2004-12-03 02:44:33 +01:00
fixed environment restoring in case of error during SP function execution (BUG#9503) #define macro improvement
2005-06-23 17:29:10 +02:00
#define reenable_binlog(A) (A)->options= tmp_disable_binlog__save_options;}
Prevent adding 'CREATE TABLE .. SELECT' query to the binary log when the insertion of new records partially failed. It would get logged because of the logic to log a partially-failed 'INSERT ... SELECT' (which can't be rolled back in non-transactional tables), but 'CREATE TABLE ... SELECT' is always rolled back on failure, even for non-transactional tables. (Bug #6682) (Original fix reimplemented after review by Serg and Guilhem.)
2004-12-03 00:05:11 +01:00
- Fix for BUG#1858 "SQL-Thread stops working when using optimize table": we change THD::system_thread from a 'bool' to a bitmap to be able to distinguish between delayed-insert threads and slave threads. - Fix for BUG#1701 "Update from multiple tables" (one line in sql_parse.cc, plus a new test rpl_multi_update.test). That's just adding an initialization.
2003-12-04 22:42:18 +01:00
added some comments
2002-03-16 09:36:27 +01:00
/*
Bug#18628 mysql-test-run: security problem(part1) - Implement --secure-file-priv=<dir> option that limits "load_file", "LOAD DATA" and "SELECT .. INTO OUTFILE" to work with files in specified dir. - Use above option for mysqld in mysql-test-run.pl
2007-02-14 14:44:34 +01:00
Used to hold information about file and file structure in exchange
added some comments
2002-03-16 09:36:27 +01:00
via non-DB file (...INTO OUTFILE..., ...LOAD DATA...)
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
XXX: We never call destructor for objects of this class.
added some comments
2002-03-16 09:36:27 +01:00
*/
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
Import changeset
2000-07-31 21:29:14 +02:00
class sql_exchange :public Sql_alloc
{
public:
char *file_name;
String *field_term,*enclosed,*line_term,*line_start,*escaped;
bool opt_enclosed;
bool dumpfile;
Extend max_allowed_packet to 2G in mysql and mysqldump (Bug #2105) Don't dump data for MRG_ISAM or MRG_MYISAM tables. (Bug #1846) Ensure that 'lower_case_table_names' is always set on case insensitive file systems. (Bug #1812) One can now configure MySQL as windows service as a normal user. (Bug #1802) Database names is now compared with lower case in ON clause when lower_case_table_names is set. (Bug #1736) IGNORE ... LINES option didn't work when used with fixed length rows. (Bug #1704) Change INSERT DELAYED ... SELECT... to INSERT .... SELECT (Bug #1983) Safety fix for service 'mysql start' (Bug #1815)
2003-12-14 05:39:52 +01:00
ulong skip_lines;
Bug#15126 character_set_database is not replicated (LOAD DATA INFILE need it) This patch fixes problem that LOAD DATA could use different character sets when loading files on master and on slave sides: - Adding replication of thd->variables.collation_database - Adding optional character set clause into LOAD DATA Note, the second way, with explicit CHARACTER SET clause should be the recommended way to load data using an alternative character set. The old way, using "SET @@character_set_database=xxx" should be gradually depricated.
2007-02-28 14:06:57 +01:00
CHARSET_INFO *cs;
Import changeset
2000-07-31 21:29:14 +02:00
sql_exchange(char *name,bool dumpfile_flag);
};
#include "log_event.h"
/*
Extension of .frm file (not yet ready for push)
2002-06-02 20:22:20 +02:00
This is used to get result from a select
Import changeset
2000-07-31 21:29:14 +02:00
*/
Making multi-table delete BETA !!
2001-07-01 12:20:53 +02:00
class JOIN;
Import changeset
2000-07-31 21:29:14 +02:00
class select_result :public Sql_alloc {
protected:
THD *thd;
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
SELECT_LEX_UNIT *unit;
Import changeset
2000-07-31 21:29:14 +02:00
public:
select_result();
virtual ~select_result() {};
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
virtual int prepare(List<Item> &list, SELECT_LEX_UNIT *u)
{
unit= u;
return 0;
}
BUG#6034 - Error code 124: Wrong medium type. Version for 5.0. Committed for merge. If the result table is one of the select tables in INSERT SELECT, we must not disable the result tables indexes before selecting. Now the preparation is split into two prepare methods. The first detects the situation and defers some preparations until the second phase.
2005-01-19 21:20:55 +01:00
virtual int prepare2(void) { return 0; }
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
/*
Because of peculiarities of prepared statements protocol
we need to know number of columns in the result set (if
there is a result set) apart from sending columns metadata.
*/
virtual uint field_count(List<Item> &fields) const
{ return fields.elements; }
Port of cursors to be pushed into 5.0 tree: - client side part is simple and may be considered stable - server side part now just joggles with THD state to save execution state and has no additional locking wisdom. Lot's of it are to be rewritten.
2004-08-03 12:32:21 +02:00
virtual bool send_fields(List<Item> &list, uint flags)=0;
Import changeset
2000-07-31 21:29:14 +02:00
virtual bool send_data(List<Item> &items)=0;
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
virtual bool initialize_tables (JOIN *join=0) { return 0; }
Fixed bug in error handling of CREATE ... SELECT More tests cases After merge fixes
2003-10-08 17:53:31 +02:00
virtual void send_error(uint errcode,const char *err);
Import changeset
2000-07-31 21:29:14 +02:00
virtual bool send_eof()=0;
A fix and a test case for Bug#24179 "select b into $var" fails with --cursor_protocol": fix a misleading error message in case of SELECT .. INTO.
2006-12-01 11:25:06 +01:00
/**
Check if this query returns a result set and therefore is allowed in
cursors and set an error message if it is not the case.
@retval FALSE success
@retval TRUE error, an error message is set
*/
virtual bool check_simple_select() const;
Import changeset
2000-07-31 21:29:14 +02:00
virtual void abort() {}
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
/*
Cleanup instance of this class for next execution of a prepared
statement/stored procedure.
*/
virtual void cleanup();
Clumsy but working fix for bug #11973 "SELECT .. INTO var_name; in trigger cause crash on update". Let us update "thd" pointer in LEX, all its units and in LEX::result before executing statement in trigger body, since triggers are associated with TABLE object and because of this can be used in different threads.
2005-08-09 22:23:56 +02:00
void set_thd(THD *thd_arg) { thd= thd_arg; }
Big patch to make embedded-server working in 5.x Now it supports queries returning several results (particularly important with the SP)
2006-01-04 11:20:28 +01:00
#ifdef EMBEDDED_LIBRARY
virtual void begin_dataset() {}
#else
void begin_dataset() {}
#endif
Import changeset
2000-07-31 21:29:14 +02:00
};
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
/*
Base class for select_result descendands which intercept and
transform result set rows. As the rows are not sent to the client,
sending of result set metadata should be suppressed as well.
*/
class select_result_interceptor: public select_result
{
public:
Fixed compiler warnings from gcc 4.0.2: - Added empty constructors and virtual destructors to many classes and structs - Removed some usage of the offsetof() macro to instead use C++ class pointers
2006-02-25 16:46:30 +01:00
select_result_interceptor() {} /* Remove gcc warning */
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
uint field_count(List<Item> &fields) const { return 0; }
bool send_fields(List<Item> &fields, uint flag) { return FALSE; }
};
Import changeset
2000-07-31 21:29:14 +02:00
class select_send :public select_result {
Fixed BUG#12379: PROCEDURE with HANDLER calling FUNCTION with error get strange result according to Monty's suggestions, fixing the SELECT behaviour on errors with SP handlers. Note that some warnings from SELECT still shows up when the handler has caught - this is an effect of another known bug (BUG#7049).
2005-09-13 15:32:42 +02:00
int status;
Import changeset
2000-07-31 21:29:14 +02:00
public:
Fixed BUG#12379: PROCEDURE with HANDLER calling FUNCTION with error get strange result according to Monty's suggestions, fixing the SELECT behaviour on errors with SP handlers. Note that some warnings from SELECT still shows up when the handler has caught - this is an effect of another known bug (BUG#7049).
2005-09-13 15:32:42 +02:00
select_send() :status(0) {}
Port of cursors to be pushed into 5.0 tree: - client side part is simple and may be considered stable - server side part now just joggles with THD state to save execution state and has no additional locking wisdom. Lot's of it are to be rewritten.
2004-08-03 12:32:21 +02:00
bool send_fields(List<Item> &list, uint flags);
Import changeset
2000-07-31 21:29:14 +02:00
bool send_data(List<Item> &items);
bool send_eof();
A fix and a test case for Bug#24179 "select b into $var" fails with --cursor_protocol": fix a misleading error message in case of SELECT .. INTO.
2006-12-01 11:25:06 +01:00
virtual bool check_simple_select() const { return FALSE; }
Fixed BUG#12379: PROCEDURE with HANDLER calling FUNCTION with error get strange result according to Monty's suggestions, fixing the SELECT behaviour on errors with SP handlers. Note that some warnings from SELECT still shows up when the handler has caught - this is an effect of another known bug (BUG#7049).
2005-09-13 15:32:42 +02:00
void abort();
Import changeset
2000-07-31 21:29:14 +02:00
};
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class select_to_file :public select_result_interceptor {
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
protected:
Import changeset
2000-07-31 21:29:14 +02:00
sql_exchange *exchange;
File file;
IO_CACHE cache;
ha_rows row_count;
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
char path[FN_REFLEN];
public:
select_to_file(sql_exchange *ex) :exchange(ex), file(-1),row_count(0L)
{ path[0]=0; }
~select_to_file();
void send_error(uint errcode,const char *err);
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
bool send_eof();
void cleanup();
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
};
class select_export :public select_to_file {
Import changeset
2000-07-31 21:29:14 +02:00
uint field_term_length;
int field_sep_char,escape_char,line_sep_char;
bool fixed_row_size;
public:
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
select_export(sql_exchange *ex) :select_to_file(ex) {}
Import changeset
2000-07-31 21:29:14 +02:00
~select_export();
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
Import changeset
2000-07-31 21:29:14 +02:00
bool send_data(List<Item> &items);
};
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
class select_dump :public select_to_file {
Import changeset
2000-07-31 21:29:14 +02:00
public:
Added SQL_SELECT::cleanup() to make it easier to reuse SQL_SELECT item's for sub selects. Needed for proper fix for bug #2479. Added 'select_to_file' class to be able to merge identical code for select_export and select_dump
2004-02-05 10:22:08 +01:00
select_dump(sql_exchange *ex) :select_to_file(ex) {}
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
Import changeset
2000-07-31 21:29:14 +02:00
bool send_data(List<Item> &items);
};
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class select_insert :public select_result_interceptor {
UNION stuff mainly.
2001-07-11 13:06:41 +02:00
public:
VIEW two TABLE_LIST copy eliminated
2004-07-16 00:15:55 +02:00
TABLE_LIST *table_list;
Import changeset
2000-07-31 21:29:14 +02:00
TABLE *table;
List<Item> *fields;
WL#3146 "less locking in auto_increment": this is a cleanup patch for our current auto_increment handling: new names for auto_increment variables in THD, new methods to manipulate them (see sql_class.h), some move into handler::, causing less backup/restore work when executing substatements. This makes the logic hopefully clearer, less work is is needed in mysql_insert(). By cleaning up, using different variables for different purposes (instead of one for 3 things...), we fix those bugs, which someone may want to fix in 5.0 too: BUG#20339 "stored procedure using LAST_INSERT_ID() does not replicate statement-based" BUG#20341 "stored function inserting into one auto_increment puts bad data in slave" BUG#19243 "wrong LAST_INSERT_ID() after ON DUPLICATE KEY UPDATE" (now if a row is updated, LAST_INSERT_ID() will return its id) and re-fixes: BUG#6880 "LAST_INSERT_ID() value changes during multi-row INSERT" (already fixed differently by Ramil in 4.1) Test of documented behaviour of mysql_insert_id() (there was no test). The behaviour changes introduced are: - LAST_INSERT_ID() now returns "the first autogenerated auto_increment value successfully inserted", instead of "the first autogenerated auto_increment value if any row was successfully inserted", see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() returns the id of the updated row if ON DUPLICATE KEY UPDATE, see auto_increment.test. Same for mysql_insert_id(), see mysql_client_test.c. - LAST_INSERT_ID() does not change if no autogenerated value was successfully inserted (it used to then be 0), see auto_increment.test. - if in INSERT SELECT no autogenerated value was successfully inserted, mysql_insert_id() now returns the id of the last inserted row (it already did this for INSERT VALUES), see mysql_client_test.c. - if INSERT SELECT uses LAST_INSERT_ID(X), mysql_insert_id() now returns X (it already did this for INSERT VALUES), see mysql_client_test.c. - NDB now behaves like other engines wrt SET INSERT_ID: with INSERT IGNORE, the id passed in SET INSERT_ID is re-used until a row succeeds; SET INSERT_ID influences not only the first row now. Additionally, when unlocking a table we check that the thread is not keeping a next_insert_id (as the table is unlocked that id is potentially out-of-date); forgetting about this next_insert_id is done in a new handler::ha_release_auto_increment(). Finally we prepare for engines capable of reserving finite-length intervals of auto_increment values: we store such intervals in THD. The next step (to be done by the replication team in 5.1) is to read those intervals from THD and actually store them in the statement-based binary log. NDB will be a good engine to test that.
2006-07-09 17:52:19 +02:00
ulonglong autoinc_value_of_last_inserted_row; // autogenerated or not
Import changeset
2000-07-31 21:29:14 +02:00
COPY_INFO info;
VIEW two TABLE_LIST copy eliminated
2004-07-16 00:15:55 +02:00
bool insert_into_view;
Import changeset
2000-07-31 21:29:14 +02:00
Merge with 4.1
2004-12-22 12:54:39 +01:00
select_insert(TABLE_LIST *table_list_par,
TABLE *table_par, List<Item> *fields_par,
WL#2274 - INSERT..SELECT..UPDATE UPDATE clause conflicts with SELECT for use of item_list field. Alter UPDATE clause to use new lex field update_list Tests included
2004-12-13 13:26:28 +01:00
List<Item> *update_fields, List<Item> *update_values,
Merge with 4.1 tree to get fix for INSERT IGNORE ... ON DUPLICATE KEY
2005-01-03 22:04:52 +01:00
enum_duplicates duplic, bool ignore);
Import changeset
2000-07-31 21:29:14 +02:00
~select_insert();
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
BUG#6034 - Error code 124: Wrong medium type. Version for 5.0. Committed for merge. If the result table is one of the select tables in INSERT SELECT, we must not disable the result tables indexes before selecting. Now the preparation is split into two prepare methods. The first detects the situation and defers some preparations until the second phase.
2005-01-19 21:20:55 +01:00
int prepare2(void);
Import changeset
2000-07-31 21:29:14 +02:00
bool send_data(List<Item> &items);
sql/sql_class.h Bug #6284 - report truncation warnings in INSERT ... SELECT only for "INSERT" part sql/sql_insert.cc Bug #6284 - report truncation warnings in INSERT ... SELECT only for "INSERT" part
2004-12-03 15:02:29 +01:00
virtual void store_values(List<Item> &values);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
virtual bool can_rollback_data() { return 0; }
Import changeset
2000-07-31 21:29:14 +02:00
void send_error(uint errcode,const char *err);
bool send_eof();
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
/* not implemented: select_insert is never re-used in prepared statements */
void cleanup();
Import changeset
2000-07-31 21:29:14 +02:00
};
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
Import changeset
2000-07-31 21:29:14 +02:00
class select_create: public select_insert {
ORDER *group;
VIEW two TABLE_LIST copy eliminated
2004-07-16 00:15:55 +02:00
TABLE_LIST *create_table;
WL#3303 (RBR: Engine-controlled logging format): Moving code to check storage engine capabilities to after tables are locked. Moving code to cache table flags so that table flags are read from the storage engine at the beginning of the statement in addition to when the storage engine is opened. To handle CREATE-SELECT, the decision function is called after the table is created and it is called with all tables that are in the select part of the statement as well as the newly created table.
2007-05-29 17:13:17 +02:00
TABLE_LIST *select_tables;
Import changeset
2000-07-31 21:29:14 +02:00
HA_CREATE_INFO *create_info;
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
Alter_info *alter_info;
Import changeset
2000-07-31 21:29:14 +02:00
Field **field;
public:
Merge mysql.com:/home/my/mysql-5.0 into mysql.com:/home/my/mysql-5.1 Merge of 'remove compiler warnings when using -Wshadow'
2007-01-27 02:46:45 +01:00
select_create (TABLE_LIST *table_arg,
VIEW two TABLE_LIST copy eliminated
2004-07-16 00:15:55 +02:00
HA_CREATE_INFO *create_info_par,
List<create_field> &fields_par,
List<Key> &keys_par,
WL#3303 (RBR: Engine-controlled logging format): Moving code to check storage engine capabilities to after tables are locked. Moving code to cache table flags so that table flags are read from the storage engine at the beginning of the statement in addition to when the storage engine is opened. To handle CREATE-SELECT, the decision function is called after the table is created and it is called with all tables that are in the select part of the statement as well as the newly created table.
2007-05-29 17:13:17 +02:00
List<Item> &select_fields,enum_duplicates duplic, bool ignore,
TABLE_LIST *select_tables_arg)
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
:select_insert (NULL, NULL, &select_fields, 0, 0, duplic, ignore),
5.1 version of a fix and test cases for bugs: Bug#4968 ""Stored procedure crash if cursor opened on altered table" Bug#6895 "Prepared Statements: ALTER TABLE DROP COLUMN does nothing" Bug#19182 "CREATE TABLE bar (m INT) SELECT n FROM foo; doesn't work from stored procedure." Bug#19733 "Repeated alter, or repeated create/drop, fails" Bug#22060 "ALTER TABLE x AUTO_INCREMENT=y in SP crashes server" Bug#24879 "Prepared Statements: CREATE TABLE (UTF8 KEY) produces a growing key length" (this bug is not fixed in 5.0) Re-execution of CREATE DATABASE, CREATE TABLE and ALTER TABLE statements in stored routines or as prepared statements caused incorrect results (and crashes in versions prior to 5.0.25). In 5.1 the problem occured only for CREATE DATABASE, CREATE TABLE SELECT and CREATE TABLE with INDEX/DATA DIRECTOY options). The problem of bugs 4968, 19733, 19282 and 6895 was that functions mysql_prepare_table, mysql_create_table and mysql_alter_table are not re-execution friendly: during their operation they modify contents of LEX (members create_info, alter_info, key_list, create_list), thus making the LEX unusable for the next execution. In particular, these functions removed processed columns and keys from create_list, key_list and drop_list. Search the code in sql_table.cc for drop_it.remove() and similar patterns to find evidence. The fix is to supply to these functions a usable copy of each of the above structures at every re-execution of an SQL statement. To simplify memory management, LEX::key_list and LEX::create_list were added to LEX::alter_info, a fresh copy of which is created for every execution. The problem of crashing bug 22060 stemmed from the fact that the above metnioned functions were not only modifying HA_CREATE_INFO structure in LEX, but also were changing it to point to areas in volatile memory of the execution memory root. The patch solves this problem by creating and using an on-stack copy of HA_CREATE_INFO in mysql_execute_command. Additionally, this patch splits the part of mysql_alter_table that analizes and rewrites information from the parser into a separate function - mysql_prepare_alter_table, in analogy with mysql_prepare_table, which is renamed to mysql_prepare_create_table.
2007-05-28 13:30:01 +02:00
create_table(table_arg),
WL#3303 (RBR: Engine-controlled logging format): Moving code to check storage engine capabilities to after tables are locked. Moving code to cache table flags so that table flags are read from the storage engine at the beginning of the statement in addition to when the storage engine is opened. To handle CREATE-SELECT, the decision function is called after the table is created and it is called with all tables that are in the select part of the statement as well as the newly created table.
2007-05-29 17:13:17 +02:00
create_info(create_info_par), select_tables(select_tables_arg)
Import changeset
2000-07-31 21:29:14 +02:00
{}
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
Merge mysql.com:/users/lthalmann/bkroot/mysql-5.1-new-rpl into mysql.com:/users/lthalmann/bk/MERGE/mysql-5.1-merge
2006-06-16 01:15:19 +02:00
WL#3023 (Use locks in a statement-like manner): Table maps are now written on aquiring locks to tables and released at the end of each logical statement.
2006-02-16 08:30:53 +01:00
void binlog_show_create_table(TABLE **tables, uint count);
sql/sql_class.h Bug #6284 - report truncation warnings in INSERT ... SELECT only for "INSERT" part sql/sql_insert.cc Bug #6284 - report truncation warnings in INSERT ... SELECT only for "INSERT" part
2004-12-03 15:02:29 +01:00
void store_values(List<Item> &values);
Prevent adding 'CREATE TABLE .. SELECT' query to the binary log when the insertion of new records partially failed. It would get logged because of the logic to log a partially-failed 'INSERT ... SELECT' (which can't be rolled back in non-transactional tables), but 'CREATE TABLE ... SELECT' is always rolled back on failure, even for non-transactional tables. (Bug #6682) (Original fix reimplemented after review by Serg and Guilhem.)
2004-12-03 00:05:11 +01:00
void send_error(uint errcode,const char *err);
Import changeset
2000-07-31 21:29:14 +02:00
bool send_eof();
void abort();
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this.
2006-06-04 17:52:22 +02:00
virtual bool can_rollback_data() { return 1; }
Improved fix for HPUX compile failure.
2006-03-24 12:24:31 +01:00
// Needed for access from local class MY_HOOKS in prepare(), since thd is proteted.
Fix for BUG#20522 "RBR: CREATE TEMPORARY TABLE SELECT writes to binlog though unneeded". It's indeed unneeded, as slave is only interested in permanent tables, and permanent tables don't depend on temporary tables when in row-based binlogging mode. And other CREATE TEMPORARY TABLE (referring no table or with LIKE) already don't write the CREATE to binlog in row-based mode.
2006-06-20 10:40:36 +02:00
const THD *get_thd(void) { return thd; }
const HA_CREATE_INFO *get_create_info() { return create_info; };
Import changeset
2000-07-31 21:29:14 +02:00
};
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
#include <myisam.h>
Fix for BUG#8218: Remove TMP_TABLE_PARAM::copy_funcs_it. TMP_TABLE_PARAM is a member of JOIN which is copied via memcpy, and List_iterator_fast TMP_TABLE_PARAM::copy_funcs_it ends up pointing to the wrong List.
2005-02-12 20:58:54 +01:00
/*
Param to create temporary tables when doing SELECT:s
NOTE
This structure is copied using memcpy as a part of JOIN.
*/
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
class TMP_TABLE_PARAM :public Sql_alloc
{
Added missing SSL library (Should be in source distribution) Fixed compiler warnings (a lot of hidden variables detected by the Forte compiler) Added a lot of 'version_xxx' strings to 'show variables' Prevent copying of TMP_TABLE_PARAM (This caused core dump bug on Solaris) Fixed problem with printing sub selects to debug log
2003-11-28 11:18:13 +01:00
private:
/* Prevent use of these (not safe because of lists and copy_field) */
TMP_TABLE_PARAM(const TMP_TABLE_PARAM &);
void operator=(TMP_TABLE_PARAM &);
public:
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
List<Item> copy_funcs;
List<Item> save_copy_funcs;
Copy_field *copy_field, *copy_field_end;
Copy_field *save_copy_field, *save_copy_field_end;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar *group_buff;
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
Item **items_to_copy; /* Fields in tmp table */
MI_COLUMNDEF *recinfo,*start_recinfo;
KEY *keyinfo;
ha_rows end_write_records;
uint field_count,sum_func_count,func_count;
uint hidden_field_count;
uint group_parts,group_length,group_null_parts;
uint quick_group;
bool using_indirect_summary_function;
Blob support for the group_concat() function. (Bug #4315: GROUP_CONCAT with ORDER BY returns strange results for TEXT fields Bug #5564: Strange behaviour with group_concat and distinct Bug #5970: group_concat doesn't print warnings)
2004-10-10 09:10:53 +02:00
/* If >0 convert all blob fields to varchar(convert_blob_length) */
uint convert_blob_length;
remove unused parts of code fix for 'show create schema_table' fix for usage schema tables in subselect 'wrong schema table charset' fix
2004-11-18 10:16:06 +01:00
CHARSET_INFO *table_charset;
Fix for bug #7217: information_schema: columns are varbinary() instead of timestamp & bug #7215: information_schema: columns are longtext instead of varchar
2005-02-23 13:15:36 +01:00
bool schema_table;
Fix for BUG#14920 Ordering aggregated result sets corrupts resultset. The cause of the bug was the use of end_write_group instead of end_write in the case when ORDER BY required a temporary table, which didn't take into account the fact that loose index scan already computes the result of MIN/MAX aggregate functions (and performs grouping). The solution is to call end_write instead of end_write_group and to add the MIN/MAX functions to the list of regular functions so that their values are inserted into the temporary table.
2005-11-30 11:52:12 +01:00
/*
True if GROUP BY and its aggregate functions are already computed
by a table access method (e.g. by loose index scan). In this case
query execution should not perform aggregation and should treat
aggregate functions as normal functions.
*/
bool precomputed_group_by;
Fixed bug#15560: GROUP_CONCAT wasn't ready for WITH ROLLUP queries The GROUP_CONCAT uses its own temporary table. When ROLLUP is present it creates the second copy of Item_func_group_concat. This copy receives the same list of arguments that original group_concat does. When the copy is set up the result_fields of functions from the argument list are reset to the temporary table of this copy. As a result of this action data from functions flow directly to the ROLLUP copy and the original group_concat functions shows wrong result. Since queries with COUNT(DISTINCT ...) use temporary tables to store the results the COUNT function they are also affected by this bug. The idea of the fix is to copy content of the result_field for the function under GROUP_CONCAT/COUNT from the first temporary table to the second one, rather than setting result_field to point to the second temporary table. To achieve this goal force_copy_fields flag is added to Item_func_group_concat and Item_sum_count_distinct classes. This flag is initialized to 0 and set to 1 into the make_unique() member function of both classes. To the TMP_TABLE_PARAM structure is modified to include the similar flag as well. The create_tmp_table() function passes that flag to create_tmp_field(). When the flag is set the create_tmp_field() function will set result_field as a source field and will not reset that result field to newly created field for Item_func_result_field and its descendants. Due to this there will be created copy func to copy data from old result_field to newly created field.
2006-03-29 21:30:34 +02:00
bool force_copy_fields;
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
TMP_TABLE_PARAM()
Fix for BUG#8218: Remove TMP_TABLE_PARAM::copy_funcs_it. TMP_TABLE_PARAM is a member of JOIN which is copied via memcpy, and List_iterator_fast TMP_TABLE_PARAM::copy_funcs_it ends up pointing to the wrong List.
2005-02-12 20:58:54 +01:00
:copy_field(0), group_parts(0),
Manual merge
2005-08-02 21:09:49 +02:00
group_length(0), group_null_parts(0), convert_blob_length(0),
Manual merge
2006-03-30 15:14:55 +02:00
schema_table(0), precomputed_group_by(0), force_copy_fields(0)
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
{}
~TMP_TABLE_PARAM()
{
cleanup();
}
Added missing SSL library (Should be in source distribution) Fixed compiler warnings (a lot of hidden variables detected by the Forte compiler) Added a lot of 'version_xxx' strings to 'show variables' Prevent copying of TMP_TABLE_PARAM (This caused core dump bug on Solaris) Fixed problem with printing sub selects to debug log
2003-11-28 11:18:13 +01:00
void init(void);
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
inline void cleanup(void)
{
if (copy_field) /* Fix for Intel compiler */
{
delete [] copy_field;
fixed joincleunup to avoid double deletin tables, and too earlyfull cleanup in case of EXPLAIN fixed cleunup of TMP_TABLE_PARAM (BUG#6406)
2004-11-15 23:16:04 +01:00
save_copy_field= copy_field= 0;
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
}
}
};
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
class select_union :public select_result_interceptor
{
Making a better fix for double released pointers and safe TMP_TABLE_PARAM. This involved moving things around in include files. All tests, including the ones with Valgrind passed.
2003-04-26 14:12:14 +02:00
TMP_TABLE_PARAM tmp_table_param;
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
public:
TABLE *table;
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
select_union() :table(0) {}
new SELECT_LEX structures used for storing global ORDER BY, global LIMIT & limit counters
2002-05-08 22:14:40 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
bool send_data(List<Item> &items);
bool send_eof();
bool flush();
A fix and a test case for Bug#6513 "Test Suite: Values inserted by using cursor is interpreted latin1 character and Bug#9819 "Cursors: Mysql Server Crash while fetching from table with 5 million records." A fix for a possible memory leak when fetching into an SP cursor in a long loop. The patch uses a common implementation of cursors in the binary protocol and in stored procedures and implements materialized cursors. For implementation details, see comments in sql_cursor.cc
2005-09-22 00:11:21 +02:00
bool create_result_table(THD *thd, List<Item> *column_types,
bool is_distinct, ulonglong options,
const char *alias);
Fix UNION New faster list iterators Change list code to be simpler and faster Optimize count(distinct) New error messages for UNION Make create_tmp_table more general to be usable by UNION
2001-08-02 05:29:50 +02:00
};
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
/* Base subselect interface class */
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class select_subselect :public select_result_interceptor
simple subselects ported to new select_lex structures
2002-05-12 22:46:42 +02:00
{
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
protected:
simple subselects ported to new select_lex structures
2002-05-12 22:46:42 +02:00
Item_subselect *item;
public:
select_subselect(Item_subselect *item);
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
bool send_data(List<Item> &items)=0;
simple subselects ported to new select_lex structures
2002-05-12 22:46:42 +02:00
bool send_eof() { return 0; };
};
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
/* Single value subselect interface class */
row IN subselects (SCRUM)
2002-12-19 20:15:09 +01:00
class select_singlerow_subselect :public select_subselect
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
{
public:
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes
2006-12-14 23:51:37 +01:00
select_singlerow_subselect(Item_subselect *item_arg)
:select_subselect(item_arg)
{}
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
bool send_data(List<Item> &items);
};
optimisation of independent ALL/ANY with aggregate function (WL#1115) (SCRUM)
2003-08-12 11:38:03 +02:00
/* used in independent ALL/ANY optimisation */
class select_max_min_finder_subselect :public select_subselect
{
Item_cache *cache;
bool (select_max_min_finder_subselect::*op)();
bool fmax;
public:
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes
2006-12-14 23:51:37 +01:00
select_max_min_finder_subselect(Item_subselect *item_arg, bool mx)
:select_subselect(item_arg), cache(0), fmax(mx)
optimisation of independent ALL/ANY with aggregate function (WL#1115) (SCRUM)
2003-08-12 11:38:03 +02:00
{}
Merge
2005-02-11 22:33:52 +01:00
void cleanup();
optimisation of independent ALL/ANY with aggregate function (WL#1115) (SCRUM)
2003-08-12 11:38:03 +02:00
bool send_data(List<Item> &items);
bool cmp_real();
bool cmp_int();
Merge
2005-02-11 22:33:52 +01:00
bool cmp_decimal();
optimisation of independent ALL/ANY with aggregate function (WL#1115) (SCRUM)
2003-08-12 11:38:03 +02:00
bool cmp_str();
};
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
/* EXISTS subselect interface class */
class select_exists_subselect :public select_subselect
{
public:
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes
2006-12-14 23:51:37 +01:00
select_exists_subselect(Item_subselect *item_arg)
:select_subselect(item_arg){}
EXISTS type of subselect more correct parameters in result creation script
2002-06-19 16:52:44 +02:00
bool send_data(List<Item> &items);
};
Import changeset
2000-07-31 21:29:14 +02:00
/* Structs used when sorting */
typedef struct st_sort_field {
Field *field; /* Field to sort */
Item *item; /* Item if not sorting fields */
uint length; /* Length of sort field */
Move handling of suffix_length from strnxfrm_bin() to filesort to ensure proper sorting of all kind of binary objects field::sort_key() now adds length last for varbinary/blob VARBINARY/BLOB is now sorted by filesort so that shorter strings comes before longer ones Fixed issues in test cases from last merge
2005-10-13 23:04:52 +02:00
uint suffix_length; /* Length suffix (0-4) */
Import changeset
2000-07-31 21:29:14 +02:00
Item_result result_type; /* Type of item */
Fixed a lot of wrong memory references as reported by valgrind Portability fixes Added new client function: mysql_get_server_version() New server help code (From Victor Vagin) Fixed wrong usage of binary() Disabled RTREE usage for now.
2003-02-12 20:55:37 +01:00
bool reverse; /* if descending sort */
bool need_strxnfrm; /* If we have to use strxnfrm() */
Import changeset
2000-07-31 21:29:14 +02:00
} SORT_FIELD;
typedef struct st_sort_buffer {
uint index; /* 0 or 1 */
uint sort_orders;
uint change_pos; /* If sort-fields changed */
char **buff;
SORT_FIELD *sortorder;
} SORT_BUFFER;
/* Structure for db & table in sql_yacc */
Fixed a lot of wrong memory references as reported by valgrind Portability fixes Added new client function: mysql_get_server_version() New server help code (From Victor Vagin) Fixed wrong usage of binary() Disabled RTREE usage for now.
2003-02-12 20:55:37 +01:00
class Table_ident :public Sql_alloc
{
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
public:
Import changeset
2000-07-31 21:29:14 +02:00
LEX_STRING db;
LEX_STRING table;
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
SELECT_LEX_UNIT *sel;
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
inline Table_ident(THD *thd, LEX_STRING db_arg, LEX_STRING table_arg,
bool force)
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
:table(table_arg), sel((SELECT_LEX_UNIT *)0)
Import changeset
2000-07-31 21:29:14 +02:00
{
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
if (!force && (thd->client_capabilities & CLIENT_NO_SCHEMA))
Import changeset
2000-07-31 21:29:14 +02:00
db.str=0;
else
db= db_arg;
}
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
inline Table_ident(LEX_STRING table_arg)
:table(table_arg), sel((SELECT_LEX_UNIT *)0)
{
db.str=0;
}
A fix and a test case for Bug#21002 "Derived table not selecting from a "real" table fails in JOINs". This is a regression caused by the fix for Bug 18444. This fix removed the assignment of empty_c_string to table->db performed in add_table_to_list, as neither me nor anyone else knew what it was there for. Now we know it and it's covered with tests: the only case when a table database name can be empty is when the table is a derived table. The fix puts the assignment back but makes it a bit more explicit. Additionally, finally drop sp.result.orig which was checked in by mistake.
2006-07-19 20:33:19 +02:00
/*
This constructor is used only for the case when we create a derived
table. A derived table has no name and doesn't belong to any database.
Later, if there was an alias specified for the table, it will be set
by add_table_to_list.
*/
foo1
2006-04-13 09:50:33 +02:00
inline Table_ident(SELECT_LEX_UNIT *s) : sel(s)
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
{
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
/* We must have a table name here as this is used with add_table_to_list */
A fix and a test case for Bug#21002 "Derived table not selecting from a "real" table fails in JOINs". This is a regression caused by the fix for Bug 18444. This fix removed the assignment of empty_c_string to table->db performed in add_table_to_list, as neither me nor anyone else knew what it was there for. Now we know it and it's covered with tests: the only case when a table database name can be empty is when the table is a derived table. The fix puts the assignment back but makes it a bit more explicit. Additionally, finally drop sp.result.orig which was checked in by mistake.
2006-07-19 20:33:19 +02:00
db.str= empty_c_string; /* a subject to casedn_str */
db.length= 0;
table.str= internal_table_name;
table.length=1;
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
}
A fix and a test case for Bug#21002 "Derived table not selecting from a "real" table fails in JOINs". This is a regression caused by the fix for Bug 18444. This fix removed the assignment of empty_c_string to table->db performed in add_table_to_list, as neither me nor anyone else knew what it was there for. Now we know it and it's covered with tests: the only case when a table database name can be empty is when the table is a derived table. The fix puts the assignment back but makes it a bit more explicit. Additionally, finally drop sp.result.orig which was checked in by mistake.
2006-07-19 20:33:19 +02:00
bool is_derived_table() const { return test(sel); }
Import changeset
2000-07-31 21:29:14 +02:00
inline void change_db(char *db_name)
new SELECT_LEX structure
2002-05-06 23:04:16 +02:00
{
db.str= db_name; db.length= (uint) strlen(db_name);
}
Import changeset
2000-07-31 21:29:14 +02:00
};
// this is needed for user_vars hash
class user_var_entry
{
public:
Fixed compiler warnings from gcc 4.0.2: - Added empty constructors and virtual destructors to many classes and structs - Removed some usage of the offsetof() macro to instead use C++ class pointers
2006-02-25 16:46:30 +01:00
user_var_entry() {} /* Remove gcc warning */
Import changeset
2000-07-31 21:29:14 +02:00
LEX_STRING name;
char *value;
Eliminate warnings noticed by VC7. This includes fixing my_mmap() on Windows to call CreateFileMapping() with correct arguments, and propogating the introduction of query_id_t to everywhere query ids are passed around. (Bug #8826)
2005-03-19 01:12:25 +01:00
ulong length;
query_id_t update_query_id, used_query_id;
Import changeset
2000-07-31 21:29:14 +02:00
Item_result type;
Bug #7498 User variable SET saves SIGNED BIGINT as UNSIGNED BIGINT - Add unsigned flag to user_var_entry, used when 'type' is INT_RESULT - Propagate unsigned flag from the query executed by Item_single_row_subselect
2006-06-09 19:35:54 +02:00
bool unsigned_flag;
Optimized code for setting user variables with := and fixed some bugs in old code (Bug #1194) Use forced close of socket to make mysqld shutdown faster when used under valgrind
2003-10-02 23:40:27 +02:00
Merge
2005-02-11 22:33:52 +01:00
double val_real(my_bool *null_value);
Optimized code for setting user variables with := and fixed some bugs in old code (Bug #1194) Use forced close of socket to make mysqld shutdown faster when used under valgrind
2003-10-02 23:40:27 +02:00
longlong val_int(my_bool *null_value);
String *val_str(my_bool *null_value, String *str, uint decimals);
Merge
2005-02-11 22:33:52 +01:00
my_decimal *val_decimal(my_bool *null_value, my_decimal *result);
New class DTCollation (SQL:2003 calls it "declared type collation") It's a combination of collation and its derivation (precedence order)
2003-06-24 12:11:07 +02:00
DTCollation collation;
Import changeset
2000-07-31 21:29:14 +02:00
};
Implementation of SUM(DISTINCT), tests cases
2003-12-19 17:04:03 +01:00
/*
Unique -- class for unique (removing of duplicates).
Puts all values to the TREE. If the tree becomes too big,
it's dumped to the file. User can request sorted values, or
just iterate through them. In the last case tree merging is performed in
memory simultaneously with iteration, so it should be ~2-3x faster.
*/
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
class Unique :public Sql_alloc
{
DYNAMIC_ARRAY file_ptrs;
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
ulong max_elements;
ulonglong max_in_memory_size;
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
IO_CACHE file;
TREE tree;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar *record_pointers;
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
bool flush();
Fixed memory allocation in Unique to not allocate too much memory
2003-06-26 06:56:55 +02:00
uint size;
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
public:
ulong elements;
Implementation of SUM(DISTINCT), tests cases
2003-12-19 17:04:03 +01:00
Unique(qsort_cmp2 comp_func, void *comp_func_fixed_arg,
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
uint size_arg, ulonglong max_in_memory_size_arg);
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
~Unique();
WL#926 "AVG(DISTINCT) and other distincts", part 2 (out of 3): clean up Item_sum_count_distinct, and deploy Unique for use with COUNT(DISTINCT) if there is no blob column in the list of DISTINCT arguments.
2005-03-15 01:46:19 +01:00
ulong elements_in_tree() { return tree.elements_in_tree; }
Implementation of SUM(DISTINCT), tests cases
2003-12-19 17:04:03 +01:00
inline bool unique_add(void *ptr)
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
{
Merge
2005-02-11 22:33:52 +01:00
DBUG_ENTER("unique_add");
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("tree %u - %lu", tree.elements_in_tree, max_elements));
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
if (tree.elements_in_tree > max_elements && flush())
Merge
2005-02-11 22:33:52 +01:00
DBUG_RETURN(1);
DBUG_RETURN(!tree_insert(&tree, ptr, 0, tree.custom_arg));
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
}
bool get(TABLE *table);
Numerous small fixes to index_merge read time estimates code
2003-12-19 22:53:14 +01:00
static double get_use_cost(uint *buffer, uint nkeys, uint key_size,
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
ulonglong max_in_memory_size);
Numerous small fixes to index_merge read time estimates code
2003-12-19 22:53:14 +01:00
inline static int get_cost_calc_buff_size(ulong nkeys, uint key_size,
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
ulonglong max_in_memory_size)
Numerous small fixes to index_merge read time estimates code
2003-12-19 22:53:14 +01:00
{
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
register ulonglong max_elems_in_tree=
Numerous small fixes to index_merge read time estimates code
2003-12-19 22:53:14 +01:00
(1 + max_in_memory_size / ALIGN_SIZE(sizeof(TREE_ELEMENT)+key_size));
Increased heap max length to > 4G for 64 bit machines Initialize key_part->type on open. This caused key_copy() to fail for bit_fields. (key_copy is used in HANDLER and opt_range)
2006-11-27 23:47:21 +01:00
return (int) (sizeof(uint)*(1 + nkeys/max_elems_in_tree));
Numerous small fixes to index_merge read time estimates code
2003-12-19 22:53:14 +01:00
}
Implementation of SUM(DISTINCT), tests cases
2003-12-19 17:04:03 +01:00
void reset();
bool walk(tree_walk_action action, void *walk_action_arg);
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
friend int unique_write_to_file(uchar* key, element_count count, Unique *unique);
friend int unique_write_to_ptrs(uchar* key, element_count count, Unique *unique);
Added Unique class to be used for duplicate removal in multi-table delete.
2001-05-23 22:47:08 +02:00
};
Enhancing both multi-table delete and division of LEX
2001-06-07 13:10:58 +02:00
Merge with 4.0.14
2003-08-11 21:44:43 +02:00
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class multi_delete :public select_result_interceptor
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
{
TABLE_LIST *delete_tables, *table_being_deleted;
Merge with 4.0.14
2003-08-11 21:44:43 +02:00
Unique **tempfiles;
sql_delete.cc, sql_update.cc, sql_class.h: Simplification of the previous fix for multi-update/delete.
2003-12-12 22:14:59 +01:00
ha_rows deleted, found;
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
uint num_of_tables;
int error;
Fix for BUG#12335 (SP replication) : New binlogging strategy for stored PROCEDUREs/FUNCTIONs. "Interleaved SPs execution is now binlogged properly, "SELECT spfunc()" is binlogged too. The known remaining issue is binlogging/replication of "a routine is deleted while it is executed" scenario.
2005-08-25 15:34:34 +02:00
bool do_delete;
/* True if at least one table we delete from is transactional */
bool transactional_tables;
/* True if at least one table we delete from is not transactional */
bool normal_tables;
bool delete_while_scanning;
Fixed bug in multiple-table-delete where some rows was not deleted
2005-05-30 19:48:40 +02:00
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
public:
Clumsy but working fix for bug #11973 "SELECT .. INTO var_name; in trigger cause crash on update". Let us update "thd" pointer in LEX, all its units and in LEX::result before executing statement in trigger body, since triggers are associated with TABLE object and because of this can be used in different threads.
2005-08-09 22:23:56 +02:00
multi_delete(TABLE_LIST *dt, uint num_of_tables);
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
~multi_delete();
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
bool send_data(List<Item> &items);
bool initialize_tables (JOIN *join);
void send_error(uint errcode,const char *err);
Fixed bug in multiple-table-delete where some rows was not deleted
2005-05-30 19:48:40 +02:00
int do_deletes();
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
bool send_eof();
};
Enhancing both multi-table delete and division of LEX
2001-06-07 13:10:58 +02:00
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class multi_update :public select_result_interceptor
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
{
support of join view updateability (WL#1809)
2004-09-15 22:42:56 +02:00
TABLE_LIST *all_tables; /* query/update command tables */
TABLE_LIST *leaves; /* list of leves of join table tree */
TABLE_LIST *update_tables, *table_being_updated;
resolve new updates group_concat
2003-04-02 15:16:19 +02:00
TABLE **tmp_tables, *main_table, *table_to_update;
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
TMP_TABLE_PARAM *tmp_table_param;
ha_rows updated, found;
List <Item> *fields, *values;
List <Item> **fields_for_table, **values_for_table;
uint table_count;
Fixed bug #28716. The result of the CHECK OPTION condition evaluation over an updated record and records of merged tables was arbitrary and dependant on the order of records in the merged tables during the execution of SELECT statement. The CHECK OPTION expression was evaluated over expired record buffers (with arbitrary data in the fields). Rowids of tables used in the CHECK OPTION expression were added to temporary table rows. The multi_update::do_updates() method was modified to restore necessary record buffers before evaluation of the CHECK OPTION condition.
2007-05-30 09:21:39 +02:00
/*
List of tables referenced in the CHECK OPTION condition of
the updated view excluding the updated table.
*/
List <TABLE> unupdated_check_opt_tables;
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
Copy_field *copy_field;
enum enum_duplicates handle_duplicates;
Fix for BUG#12335 (SP replication) : New binlogging strategy for stored PROCEDUREs/FUNCTIONs. "Interleaved SPs execution is now binlogged properly, "SELECT spfunc()" is binlogged too. The known remaining issue is binlogging/replication of "a routine is deleted while it is executed" scenario.
2005-08-25 15:34:34 +02:00
bool do_update, trans_safe;
/* True if the update operation has made a change in a transactional table */
bool transactional_tables;
bool ignore;
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
public:
Clumsy but working fix for bug #11973 "SELECT .. INTO var_name; in trigger cause crash on update". Let us update "thd" pointer in LEX, all its units and in LEX::result before executing statement in trigger body, since triggers are associated with TABLE object and because of this can be used in different threads.
2005-08-09 22:23:56 +02:00
multi_update(TABLE_LIST *ut, TABLE_LIST *leaves_list,
support of join view updateability (WL#1809)
2004-09-15 22:42:56 +02:00
List<Item> *fields, List<Item> *values,
Merge with 4.1 tree to get fix for INSERT IGNORE ... ON DUPLICATE KEY
2005-01-03 22:04:52 +01:00
enum_duplicates handle_duplicates, bool ignore);
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
~multi_update();
Merge with 4.0.6
2002-12-05 18:38:42 +01:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
New multi-table-update code New (simpler) internal timestamp handling. More debuging to heap tables. Small cleanups to multi-table-delete false -> 0 and true -> 1 (We should use TRUE and FALSE)
2002-11-29 15:40:18 +01:00
bool send_data(List<Item> &items);
bool initialize_tables (JOIN *join);
void send_error(uint errcode,const char *err);
int do_updates (bool from_send_error);
bool send_eof();
};
This is a large push. Included are : * multi-table updates * new paid feature for limiting number of queries per hour for users * optional syntax for multi-table deletes * optimization for SQL_CALC_FOUND_ROWS * a small addition for CREATE .. SELECT that will be of future use I know that all this will require many additions to documentation, which I have not done, but I am at Arjen's disposal to help him document all this.
2001-12-26 15:49:10 +01:00
SELECT ... INTO local_vars ...; For Stored Procedures
2003-01-18 17:21:13 +01:00
class my_var : public Sql_alloc {
public:
LEX_STRING s;
A name of macro is fixed.
2005-11-23 11:26:07 +01:00
#ifndef DBUG_OFF
Fix for BUG#13549 "Server crash with nested stored procedures if inner routine has more local variables than outer one, and one of its last variables was used as argument to NOT operator". THD::spcont was non-0 when we were parsing stored routine/trigger definition during execution of another stored routine. This confused methods of Item_splocal and forced them use wrong runtime context. Fix ensures that we always have THD::spcont equal to zero during routine/trigger body parsing. This also allows to avoid problems with errors which occur during parsing and SQL exception handlers.
2005-11-22 23:50:37 +01:00
/*
Routine to which this Item_splocal belongs. Used for checking if correct
runtime context is used for variable handling.
*/
Patch for WL#2894: Make stored routine variables work according to the standard. The idea is to use Field-classes to implement stored routines variables. Also, we should provide facade to Item-hierarchy by Item_field class (it is necessary, since SRVs take part in expressions). The patch fixes the following bugs: - BUG#8702: Stored Procedures: No Error/Warning shown for inappropriate data type matching; - BUG#8768: Functions: For any unsigned data type, -ve values can be passed and returned; - BUG#8769: Functions: For Int datatypes, out of range values can be passed and returned; - BUG#9078: STORED PROCDURE: Decimal digits are not displayed when we use DECIMAL datatype; - BUG#9572: Stored procedures: variable type declarations ignored; - BUG#12903: upper function does not work inside a function; - BUG#13705: parameters to stored procedures are not verified; - BUG#13808: ENUM type stored procedure parameter accepts non-enumerated data; - BUG#13909: Varchar Stored Procedure Parameter always BINARY string (ignores CHARACTER SET); - BUG#14161: Stored procedure cannot retrieve bigint unsigned; - BUG#14188: BINARY variables have no 0x00 padding; - BUG#15148: Stored procedure variables accept non-scalar values;
2005-12-07 15:01:17 +01:00
sp_head *sp;
Fix for BUG#13549 "Server crash with nested stored procedures if inner routine has more local variables than outer one, and one of its last variables was used as argument to NOT operator". THD::spcont was non-0 when we were parsing stored routine/trigger definition during execution of another stored routine. This confused methods of Item_splocal and forced them use wrong runtime context. Fix ensures that we always have THD::spcont equal to zero during routine/trigger body parsing. This also allows to avoid problems with errors which occur during parsing and SQL exception handlers.
2005-11-22 23:50:37 +01:00
#endif
SELECT ... INTO local_vars ...; For Stored Procedures
2003-01-18 17:21:13 +01:00
bool local;
uint offset;
Fix for BUG#1495: Evaluate items before setting a local variable with SELECT INTO. Also copy and restore order_list and group_list for selects in SPs.
2003-10-14 12:59:28 +02:00
enum_field_types type;
my_var (LEX_STRING& j, bool i, uint o, enum_field_types t)
:s(j), local(i), offset(o), type(t)
{}
SELECT ... INTO local_vars ...; For Stored Procedures
2003-01-18 17:21:13 +01:00
~my_var() {}
};
This is a large push. Included are : * multi-table updates * new paid feature for limiting number of queries per hour for users * optional syntax for multi-table deletes * optimization for SQL_CALC_FOUND_ROWS * a small addition for CREATE .. SELECT that will be of future use I know that all this will require many additions to documentation, which I have not done, but I am at Arjen's disposal to help him document all this.
2001-12-26 15:49:10 +01:00
A fix and test case for bug#6059 "mysql_stmt_field_count returns positive numbers when no resultset is available": when sending result set metadata we need to use virtual select_result::send_fields, and not address protocol directly, because select_result descendents may intercept result set (it's the case for example for SELECT INTO OUTFILE).
2004-10-21 16:33:53 +02:00
class select_dumpvar :public select_result_interceptor {
SCRUM task
2002-10-11 20:49:10 +02:00
ha_rows row_count;
public:
SELECT ... INTO local_vars ...; For Stored Procedures
2003-01-18 17:21:13 +01:00
List<my_var> var_list;
Bug#20836 Selecting into variables results in wrong results being returned This error is displayed anytime the SELECT statement needs a temp table to return correct results because the object (select_dumpvar) that represents variables named in the INTO clause stored the results before the temp table was considered. The problem was fixed by creating the necessary Item_func_set_user_var objects once the correct data is ready.
2006-11-28 23:21:39 +01:00
select_dumpvar() { var_list.empty(); row_count= 0;}
SCRUM task
2002-10-11 20:49:10 +02:00
~select_dumpvar() {}
some fixes for SELECT INTO @vars ..
2002-10-16 15:55:08 +02:00
int prepare(List<Item> &list, SELECT_LEX_UNIT *u);
SCRUM task
2002-10-11 20:49:10 +02:00
bool send_data(List<Item> &items);
bool send_eof();
A fix and a test case for Bug#24179 "select b into $var" fails with --cursor_protocol": fix a misleading error message in case of SELECT .. INTO.
2006-12-01 11:25:06 +01:00
virtual bool check_simple_select() const;
Fix for Bug#5034 "prepared "select 1 into @arg15", second execute crashes server": we were deleting lex->result after each execute, but prepared statements assumed that it's left intact. The fix adds cleanup() method to select_result hierarchy, so that result objects can be reused. Plus we now need to delete result objects more wisely.
2004-08-24 18:17:11 +02:00
void cleanup();
SCRUM task
2002-10-11 20:49:10 +02:00
};
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
SHOW STATUS does not anymore change local status variables (except com_show_status). Global status variables are still updated. SHOW STATUS are not anymore put in slow query log because of no index usage. Implemntation done by removing orig_sql_command and moving logic of SHOW STATUS to mysql_excute_command() This simplifies code and allows us to remove some if statements all over the code. Upgraded uc_update_queries[] to sql_command_flags and added more bitmaps to better categorize commands. This allowed some overall simplifaction when testing sql_command. Fixes bugs: Bug#10210: running SHOW STATUS increments counters it shouldn't Bug#19764: SHOW commands end up in the slow log as table scans
2006-06-20 12:20:32 +02:00
/* Bits in sql_command_flags */
#define CF_CHANGES_DATA 1
#define CF_HAS_ROW_COUNT 2
#define CF_STATUS_COMMAND 4
#define CF_SHOW_TABLE_COMMAND 8
Added per thread status variables, after review, patch v1.1.
2004-09-13 15:48:01 +02:00
/* Functions in sql_class.cc */
void add_to_status(STATUS_VAR *to_var, STATUS_VAR *from_var);
SHOW STATUS does not anymore change local status variables (except com_show_status). Global status variables are still updated. SHOW STATUS are not anymore put in slow query log because of no index usage. Implemntation done by removing orig_sql_command and moving logic of SHOW STATUS to mysql_excute_command() This simplifies code and allows us to remove some if statements all over the code. Upgraded uc_update_queries[] to sql_command_flags and added more bitmaps to better categorize commands. This allowed some overall simplifaction when testing sql_command. Fixes bugs: Bug#10210: running SHOW STATUS increments counters it shouldn't Bug#19764: SHOW commands end up in the slow log as table scans
2006-06-20 12:20:32 +02:00
void add_diff_to_status(STATUS_VAR *to_var, STATUS_VAR *from_var,
STATUS_VAR *dec_var);
Fix the tree: plugin-related changes, necessary for CSV storage engine to be statically compiled. Derived from antony's patch.
2006-01-19 22:40:56 +01:00
#endif /* MYSQL_SERVER */
Reference in a new issue Copy permalink