fil0fil.c 104 KB
Newer Older
1
/******************************************************
unknown's avatar
unknown committed
2
The tablespace memory cache
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

(c) 1995 Innobase Oy

Created 10/25/1995 Heikki Tuuri
*******************************************************/

#include "fil0fil.h"

#include "mem0mem.h"
#include "sync0sync.h"
#include "hash0hash.h"
#include "os0file.h"
#include "os0sync.h"
#include "mach0data.h"
#include "ibuf0ibuf.h"
#include "buf0buf.h"
unknown's avatar
unknown committed
19 20
#include "buf0flu.h"
#include "buf0lru.h"
21 22 23
#include "log0log.h"
#include "log0recv.h"
#include "fsp0fsp.h"
24
#include "srv0srv.h"
unknown's avatar
unknown committed
25
#include "srv0start.h"
unknown's avatar
unknown committed
26 27
#include "mtr0mtr.h"
#include "mtr0log.h"
28

unknown's avatar
unknown committed
29
	 
30
/*
unknown's avatar
unknown committed
31 32
		IMPLEMENTATION OF THE TABLESPACE MEMORY CACHE
		=============================================
33

unknown's avatar
unknown committed
34
The tablespace cache is responsible for providing fast read/write access to
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85
tablespaces and logs of the database. File creation and deletion is done
in other modules which know more of the logic of the operation, however.

A tablespace consists of a chain of files. The size of the files does not
have to be divisible by the database block size, because we may just leave
the last incomplete block unused. When a new file is appended to the
tablespace, the maximum size of the file is also specified. At the moment,
we think that it is best to extend the file to its maximum size already at
the creation of the file, because then we can avoid dynamically extending
the file when more space is needed for the tablespace.

A block's position in the tablespace is specified with a 32-bit unsigned
integer. The files in the chain are thought to be catenated, and the block
corresponding to an address n is the nth block in the catenated file (where
the first block is named the 0th block, and the incomplete block fragments
at the end of files are not taken into account). A tablespace can be extended
by appending a new file at the end of the chain.

Our tablespace concept is similar to the one of Oracle.

To acquire more speed in disk transfers, a technique called disk striping is
sometimes used. This means that logical block addresses are divided in a
round-robin fashion across several disks. Windows NT supports disk striping,
so there we do not need to support it in the database. Disk striping is
implemented in hardware in RAID disks. We conclude that it is not necessary
to implement it in the database. Oracle 7 does not support disk striping,
either.

Another trick used at some database sites is replacing tablespace files by
raw disks, that is, the whole physical disk drive, or a partition of it, is
opened as a single file, and it is accessed through byte offsets calculated
from the start of the disk or the partition. This is recommended in some
books on database tuning to achieve more speed in i/o. Using raw disk
certainly prevents the OS from fragmenting disk space, but it is not clear
if it really adds speed. We measured on the Pentium 100 MHz + NT + NTFS file
system + EIDE Conner disk only a negligible difference in speed when reading
from a file, versus reading from a raw disk. 

To have fast access to a tablespace or a log file, we put the data structures
to a hash table. Each tablespace and log file is given an unique 32-bit
identifier.

Some operating systems do not support many open files at the same time,
though NT seems to tolerate at least 900 open files. Therefore, we put the
open files in an LRU-list. If we need to open another file, we may close the
file at the end of the LRU-list. When an i/o-operation is pending on a file,
the file cannot be closed. We take the file nodes with pending i/o-operations
out of the LRU-list and keep a count of pending operations. When an operation
completes, we decrement the count and return the file node to the LRU-list if
the count drops to zero. */

unknown's avatar
unknown committed
86 87 88 89
/* When mysqld is run, the default directory "." is the mysqld datadir,
but in ibbackup we must set it explicitly */
char*	fil_path_to_mysql_datadir	= (char*)".";

90 91 92
ulint	fil_n_pending_log_flushes		= 0;
ulint	fil_n_pending_tablespace_flushes	= 0;

93 94 95
/* Null file address */
fil_addr_t	fil_addr_null = {FIL_NULL, 0};

unknown's avatar
unknown committed
96
/* File node of a tablespace or the log data space */
97 98
typedef	struct fil_node_struct	fil_node_t;
struct fil_node_struct {
unknown's avatar
unknown committed
99 100 101
	fil_space_t*	space;	/* backpointer to the space where this node
				belongs */
	char*		name;	/* path to the file */
102 103
	ibool		open;	/* TRUE if file open */
	os_file_t	handle;	/* OS handle to the file, if file open */
unknown's avatar
unknown committed
104 105 106 107 108
	ibool		is_raw_disk;/* TRUE if the 'file' is actually a raw
				device or a raw disk partition */
	ulint		size;	/* size of the file in database pages, 0 if
				not known yet; the possible last incomplete
				megabyte is ignored if space == 0 */
109
	ulint		n_pending;
unknown's avatar
unknown committed
110 111 112 113 114 115 116 117 118 119 120
				/* count of pending i/o's on this file;
				closing of the file is not allowed if
				this is > 0 */
	ulint		n_pending_flushes;
				/* count of pending flushes on this file;
				closing of the file is not allowed if
				this is > 0 */	
	ib_longlong	modification_counter;/* when we write to the file we
				increment this by one */
	ib_longlong	flush_counter;/* up to what modification_counter value
				we have flushed the modifications to disk */
121 122 123 124 125 126 127 128 129
	UT_LIST_NODE_T(fil_node_t) chain;
				/* link field for the file chain */
	UT_LIST_NODE_T(fil_node_t) LRU;
				/* link field for the LRU list */
	ulint		magic_n;
};

#define	FIL_NODE_MAGIC_N	89389

unknown's avatar
unknown committed
130
/* Tablespace or log data space: let us call them by a common name space */
131
struct fil_space_struct {
unknown's avatar
unknown committed
132 133
	char*		name;	/* space name = the path to the first file in
				it */
134
	ulint		id;	/* space id */
unknown's avatar
unknown committed
135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
	ib_longlong	tablespace_version;
				/* in DISCARD/IMPORT this timestamp is used to
				check if we should ignore an insert buffer
				merge request for a page because it actually
				was for the previous incarnation of the
				space */
	ibool		mark;	/* this is set to TRUE at database startup if
				the space corresponds to a table in the InnoDB
				data dictionary; so we can print a warning of
				orphaned tablespaces */
	ibool		stop_ios;/* TRUE if we want to rename the .ibd file of
				tablespace and want to stop temporarily
				posting of new i/o requests on the file */
	ibool		stop_ibuf_merges;
				/* we set this TRUE when we start deleting a
				single-table tablespace */
	ibool		is_being_deleted;
				/* this is set to TRUE when we start
				deleting a single-table tablespace and its
				file; when this flag is set no further i/o
				or flush requests can be placed on this space,
				though there may be such requests still being
				processed on this space */
158 159 160
	ulint		purpose;/* FIL_TABLESPACE, FIL_LOG, or FIL_ARCH_LOG */
	UT_LIST_BASE_NODE_T(fil_node_t) chain;
				/* base node for the file chain */
unknown's avatar
unknown committed
161 162
	ulint		size;	/* space size in pages; 0 if a single-table
				tablespace whose size we do not know yet */
163 164 165
	ulint		n_reserved_extents;
				/* number of reserved free extents for
				ongoing operations like B-tree page split */
unknown's avatar
unknown committed
166 167 168 169 170 171 172 173
	ulint		n_pending_flushes; /* this is > 0 when flushing
				the tablespace to disk; dropping of the
				tablespace is forbidden if this is > 0 */
	ulint		n_pending_ibuf_merges;/* this is > 0 when merging
				insert buffer entries to a page so that we
				may need to access the ibuf bitmap page in the
				tablespade: dropping of the tablespace is
				forbidden if this is > 0 */
174
	hash_node_t	hash; 	/* hash chain node */
unknown's avatar
unknown committed
175
	hash_node_t	name_hash;/* hash chain the name_hash table */
176 177 178 179 180 181 182 183 184 185 186
	rw_lock_t	latch;	/* latch protecting the file space storage
				allocation */
	UT_LIST_NODE_T(fil_space_t) space_list;
				/* list of all spaces */
	ibuf_data_t*	ibuf_data;
				/* insert buffer data */
	ulint		magic_n;
};

#define	FIL_SPACE_MAGIC_N	89472

unknown's avatar
unknown committed
187 188 189
/* The tablespace memory cache; also the totality of logs = the log data space,
is stored here; below we talk about tablespaces, but also the ib_logfiles
form a 'space' and it is handled here */
190 191 192

typedef	struct fil_system_struct	fil_system_t;
struct fil_system_struct {
unknown's avatar
unknown committed
193
	mutex_t		mutex;		/* The mutex protecting the cache */
194
	hash_table_t*	spaces;		/* The hash table of spaces in the
unknown's avatar
unknown committed
195 196 197 198
					system; they are hashed on the space
					id */
	hash_table_t*	name_hash;	/* hash table based on the space
					name */
199 200
	UT_LIST_BASE_NODE_T(fil_node_t) LRU;
					/* base node for the LRU list of the
unknown's avatar
unknown committed
201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
					most recently used open files with no
					pending i/o's; if we start an i/o on
					the file, we first remove it from this
					list, and return it to the start of
					the list when the i/o ends;
					log files and the system tablespace are
					not put to this list: they are opened
					after the startup, and kept open until
					shutdown */
	ulint		n_open;		/* number of files currently open */
	ulint		max_n_open;	/* n_open is not allowed to exceed
					this */
	ib_longlong	modification_counter;/* when we write to a file we
					increment this by one */
	ulint		max_assigned_id;/* maximum space id in the existing
					tables, or assigned during the time
					mysqld has been up; at an InnoDB
					startup we scan the data dictionary
					and set here the maximum of the
					space id's of the tables there */
	ib_longlong	tablespace_version;
					/* a counter which is incremented for
					every space object memory creation;
					every space mem object gets a
					'timestamp' from this; in DISCARD/
					IMPORT this is used to check if we
					should ignore an insert buffer merge
					request */
229 230 231 232
	UT_LIST_BASE_NODE_T(fil_space_t) space_list;
					/* list of all file spaces */
};

unknown's avatar
unknown committed
233 234
/* The tablespace memory cache. This variable is NULL before the module is
initialized. */
235 236
fil_system_t*	fil_system	= NULL;

unknown's avatar
unknown committed
237 238
/* The tablespace memory cache hash table size */
#define	FIL_SYSTEM_HASH_SIZE	50 /* TODO: make bigger! */
239 240


unknown's avatar
unknown committed
241 242
/************************************************************************
NOTE: you must call fil_mutex_enter_and_prepare_for_io() first!
243

unknown's avatar
unknown committed
244 245 246 247 248
Prepares a file node for i/o. Opens the file if it is closed. Updates the
pending i/o's field in the node and the system appropriately. Takes the node
off the LRU list if it is in the LRU list. The caller must hold the fil_sys
mutex. */
static
249
void
unknown's avatar
unknown committed
250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
fil_node_prepare_for_io(
/*====================*/
	fil_node_t*	node,	/* in: file node */
	fil_system_t*	system,	/* in: tablespace memory cache */
	fil_space_t*	space);	/* in: space */
/************************************************************************
Updates the data structures when an i/o operation finishes. Updates the
pending i/o's field in the node appropriately. */
static
void
fil_node_complete_io(
/*=================*/
	fil_node_t*	node,	/* in: file node */
	fil_system_t*	system,	/* in: tablespace memory cache */
	ulint		type);	/* in: OS_FILE_WRITE or OS_FILE_READ; marks
				the node as modified if
				type == OS_FILE_WRITE */
unknown's avatar
unknown committed
267 268 269 270 271 272 273 274 275 276 277
/***********************************************************************
Checks if a single-table tablespace for a given table name exists in the
tablespace memory cache. */
static
ulint
fil_get_space_id_for_table(
/*=======================*/
				/* out: space id, ULINT_UNDEFINED if not
				found */
	char*	name);		/* in: table name in the standard
				'databasename/tablename' format */
278 279


unknown's avatar
unknown committed
280 281
/***********************************************************************
Returns the version number of a tablespace, -1 if not found. */
282

unknown's avatar
unknown committed
283 284 285 286 287 288 289 290 291 292
ib_longlong
fil_space_get_version(
/*==================*/
			/* out: version number, -1 if the tablespace does not
			exist in the memory cache */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;
	ib_longlong	version		= -1;
293

unknown's avatar
unknown committed
294
	ut_ad(system);
295

unknown's avatar
unknown committed
296
	mutex_enter(&(system->mutex));
297

unknown's avatar
unknown committed
298
	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);
299

unknown's avatar
unknown committed
300 301
	if (space) {
		version = space->tablespace_version;
302 303
	}

unknown's avatar
unknown committed
304
	mutex_exit(&(system->mutex));
305

unknown's avatar
unknown committed
306
	return(version);
307 308 309 310 311 312 313 314 315 316 317 318
}

/***********************************************************************
Returns the latch of a file space. */

rw_lock_t*
fil_space_get_latch(
/*================*/
			/* out: latch protecting storage allocation */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
unknown's avatar
unknown committed
319
	fil_space_t*	space;
320 321 322 323 324 325 326

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
327 328
	ut_a(space);

329 330 331 332 333 334 335 336 337 338 339 340 341 342 343
	mutex_exit(&(system->mutex));

	return(&(space->latch));
}

/***********************************************************************
Returns the type of a file space. */

ulint
fil_space_get_type(
/*===============*/
			/* out: FIL_TABLESPACE or FIL_LOG */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
unknown's avatar
unknown committed
344
	fil_space_t*	space;
345 346 347 348 349 350 351

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
352 353
	ut_a(space);

354 355 356 357 358 359 360 361 362 363 364 365 366 367
	mutex_exit(&(system->mutex));

	return(space->purpose);
}

/***********************************************************************
Returns the ibuf data of a file space. */

ibuf_data_t*
fil_space_get_ibuf_data(
/*====================*/
			/* out: ibuf data for this space */
	ulint	id)	/* in: space id */
{
unknown's avatar
unknown committed
368
	fil_system_t*	system		= fil_system;
369 370 371 372
	fil_space_t*	space;

	ut_ad(system);

unknown's avatar
unknown committed
373 374
	ut_a(id == 0);

375 376 377 378 379 380
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
381 382
	ut_a(space);

383 384 385 386 387 388 389 390 391 392 393 394
	return(space->ibuf_data);
}

/***********************************************************************
Appends a new file to the chain of files of a space. File must be closed. */

void
fil_node_create(
/*============*/
	char*	name,	/* in: file name (file must be closed) */
	ulint	size,	/* in: file size in database blocks, rounded downwards
			to an integer */
unknown's avatar
unknown committed
395 396
	ulint	id,	/* in: space id where to append */
	ibool	is_raw)	/* in: TRUE if a raw device or a raw disk partition */
397
{
unknown's avatar
unknown committed
398
	fil_system_t*	system	= fil_system;
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
	fil_node_t*	node;
	fil_space_t*	space;
	char*		name2;

	ut_a(system);
	ut_a(name);

	mutex_enter(&(system->mutex));

	node = mem_alloc(sizeof(fil_node_t));

	name2 = mem_alloc(ut_strlen(name) + 1);

	ut_strcpy(name2, name);

	node->name = name2;
	node->open = FALSE;
unknown's avatar
unknown committed
416 417 418 419

	ut_a(!is_raw || srv_start_raw_disk_in_use);

	node->is_raw_disk = is_raw;
420 421 422
	node->size = size;
	node->magic_n = FIL_NODE_MAGIC_N;
	node->n_pending = 0;
unknown's avatar
unknown committed
423
	node->n_pending_flushes = 0;
424

unknown's avatar
unknown committed
425 426
	node->modification_counter = 0;
	node->flush_counter = 0;
427 428 429
	
	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
430 431 432 433 434 435 436 437 438 439 440 441 442 443
	if (!space) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Error: Could not find tablespace %lu for\n"
"InnoDB: file %s from the tablespace memory cache.\n", id, name);
		mem_free(name2);

		mem_free(node);

		mutex_exit(&(system->mutex));

		return;
	}

444 445
	space->size += size;

unknown's avatar
unknown committed
446 447
	node->space = space;

448 449 450 451 452
	UT_LIST_ADD_LAST(chain, space->chain, node);
				
	mutex_exit(&(system->mutex));
}

unknown's avatar
unknown committed
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
/************************************************************************
Opens a the file of a node of a tablespace. The caller must own the fil_system
mutex. */
static
void
fil_node_open_file(
/*===============*/
	fil_node_t*	node,	/* in: file node */
	fil_system_t*	system,	/* in: tablespace memory cache */
	fil_space_t*	space)	/* in: space */
{
	ib_longlong	size_bytes;
	ulint		size_low;
	ulint		size_high;
	ibool		ret;

	ut_ad(mutex_own(&(system->mutex)));

	ut_a(node->n_pending == 0);
	ut_a(node->open == FALSE);

	/* printf("Opening file %s\n", node->name); */

	if (space->purpose == FIL_LOG) {	
		node->handle = os_file_create(node->name, OS_FILE_OPEN,
					OS_FILE_AIO, OS_LOG_FILE, &ret);
	} else if (node->is_raw_disk) {
		node->handle = os_file_create(node->name,
				        OS_FILE_OPEN_RAW,
					OS_FILE_AIO, OS_DATA_FILE, &ret);
	} else {
		node->handle = os_file_create(node->name, OS_FILE_OPEN,
					OS_FILE_AIO, OS_DATA_FILE, &ret);
	}
		
	ut_a(ret);
		
	node->open = TRUE;

	system->n_open++;

	if (node->size == 0) {
unknown's avatar
unknown committed
495 496 497 498 499 500 501 502
		os_file_get_size(node->handle, &size_low, &size_high);

		size_bytes = (((ib_longlong)size_high) << 32)
				     		+ (ib_longlong)size_low;
#ifdef UNIV_HOTBACKUP
		node->size = (ulint) (size_bytes / UNIV_PAGE_SIZE);

#else
unknown's avatar
unknown committed
503 504 505 506 507 508 509 510 511 512 513
		/* It must be a single-table tablespace and we do not know the
		size of the file yet */

		ut_a(space->id != 0);

		if (size_bytes >= FSP_EXTENT_SIZE * UNIV_PAGE_SIZE) {
			node->size = (ulint) ((size_bytes / (1024 * 1024))
					   * ((1024 * 1024) / UNIV_PAGE_SIZE));
		} else {
			node->size = (ulint) (size_bytes / UNIV_PAGE_SIZE);
		}
unknown's avatar
unknown committed
514 515
#endif
		space->size += node->size;
unknown's avatar
unknown committed
516 517 518 519 520 521 522 523
	}

	if (space->purpose == FIL_TABLESPACE && space->id != 0) {
		/* Put the node to the LRU list */
		UT_LIST_ADD_FIRST(LRU, system->LRU, node);
	}
}

524 525 526 527
/**************************************************************************
Closes a file. */
static
void
unknown's avatar
unknown committed
528 529
fil_node_close_file(
/*================*/
530
	fil_node_t*	node,	/* in: file node */
unknown's avatar
unknown committed
531
	fil_system_t*	system)	/* in: tablespace memory cache */
532 533 534 535 536 537 538
{
	ibool	ret;

	ut_ad(node && system);
	ut_ad(mutex_own(&(system->mutex)));
	ut_a(node->open);
	ut_a(node->n_pending == 0);
unknown's avatar
unknown committed
539
	ut_a(node->n_pending_flushes == 0);
unknown's avatar
unknown committed
540
	ut_a(node->modification_counter == node->flush_counter);
541 542 543 544

	ret = os_file_close(node->handle);
	ut_a(ret);

unknown's avatar
unknown committed
545 546
	/* printf("Closing file %s\n", node->name); */

547
	node->open = FALSE;
unknown's avatar
unknown committed
548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590
	ut_a(system->n_open > 0);
	system->n_open--;

	if (node->space->purpose == FIL_TABLESPACE && node->space->id != 0) {
		ut_a(UT_LIST_GET_LEN(system->LRU) > 0);

		/* The node is in the LRU list, remove it */
		UT_LIST_REMOVE(LRU, system->LRU, node);
	}
}

/************************************************************************
Tries to close a file in the LRU list. The caller must hold the fil_sys
mutex. */
static
ibool
fil_try_to_close_file_in_LRU(
/*=========================*/
				/* out: TRUE if success, FALSE if should retry
				later; since i/o's generally complete in < 
				100 ms, and as InnoDB writes at most 128 pages
				from the buffer pool in a batch, and then
				immediately flushes the files, there is a good
				chance that the next time we find a suitable
				node from the LRU list */
	ibool	print_info)	/* in: if TRUE, prints information why it
				cannot close a file */
{
	fil_system_t*	system		= fil_system;
	fil_node_t*	node;

	ut_ad(mutex_own(&(system->mutex)));

	node = UT_LIST_GET_LAST(system->LRU);

	if (print_info) {
		fprintf(stderr,
"InnoDB: fil_sys open file LRU len %lu\n", UT_LIST_GET_LEN(system->LRU));
	}

	while (node != NULL) {
		if (node->modification_counter == node->flush_counter
		    && node->n_pending_flushes == 0) {
591

unknown's avatar
unknown committed
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714
			fil_node_close_file(node, system);
			
			return(TRUE);
		}
		
		if (print_info && node->n_pending_flushes > 0) {
			fprintf(stderr,
"InnoDB: cannot close file %s, because n_pending_flushes %lu\n", node->name,
				       node->n_pending_flushes);
		}

		if (print_info
		    && node->modification_counter != node->flush_counter) {
			fprintf(stderr,
"InnoDB: cannot close file %s, because mod_count %lld != fl_count %lld\n",
				node->name, node->modification_counter,
				node->flush_counter);
		}

		node = UT_LIST_GET_PREV(LRU, node);
	}

	return(FALSE);
}

/***********************************************************************
Reserves the fil_system mutex and tries to make sure we can open at least one
file while holding it. This should be called before calling
fil_node_prepare_for_io(), because that function may need to open a file. */
static
void
fil_mutex_enter_and_prepare_for_io(
/*===============================*/
	ulint	space_id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;
	ibool		success;
	ibool		print_info	= FALSE;
	ulint		count		= 0;
	ulint		count2		= 0;

	ut_ad(!mutex_own(&(system->mutex)));
retry:
	mutex_enter(&(system->mutex));

	if (space_id == 0 || space_id >= SRV_LOG_SPACE_FIRST_ID) {
		/* We keep log files and system tablespace files always open;
		this is important in preventing deadlocks in this module, as
		a page read completion often performs another read from the
		insert buffer. The insert buffer is in tablespace 0, and we
		cannot end up waiting in this function. */

		return;
	}

	if (system->n_open < system->max_n_open) {

		return;
	}

	HASH_SEARCH(hash, system->spaces, space_id, space,
							space->id == space_id);
	if (space != NULL && space->stop_ios) {
		/* We are going to do a rename file and want to stop new i/o's
		for a while */

		if (count2 > 20000) {
			fprintf(stderr,
"InnoDB: Warning: tablespace %s has i/o ops stopped for a long time %lu\n",
						    space->name, count2);
		}

		mutex_exit(&(system->mutex));

		os_thread_sleep(20000);

		count2++;

		goto retry;
	}

	/* If the file is already open, no need to do anything; if the space
	does not exist, we handle the situation in the function which called
	this function */

	if (!space || UT_LIST_GET_FIRST(space->chain)->open) {

		return;
	}

	if (count > 1) {
		print_info = TRUE;
	}

	/* Too many files are open, try to close some */
close_more:
	success = fil_try_to_close_file_in_LRU(print_info);

	if (success && system->n_open >= system->max_n_open) {

		goto close_more;
	}

	if (system->n_open < system->max_n_open) {
		/* Ok */

		return;
	}

	if (count >= 2) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Warning: too many (%lu) files stay open while the maximum\n"
"InnoDB: allowed value would be %lu.\n"
"InnoDB: You may need to raise the value of innodb_max_files_open in\n"
"InnoDB: my.cnf.\n", system->n_open, system->max_n_open);

		return;
	}

	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
715
#ifndef UNIV_HOTBACKUP
unknown's avatar
unknown committed
716 717 718 719 720
	/* Wake the i/o-handler threads to make sure pending i/o's are
	performed */
	os_aio_simulated_wake_handler_threads();

	os_thread_sleep(20000);
unknown's avatar
unknown committed
721
#endif
unknown's avatar
unknown committed
722 723 724 725 726 727 728 729
	/* Flush tablespaces so that we can close modified files in the LRU
	list */

	fil_flush_file_spaces(FIL_TABLESPACE);		

	count++;

	goto retry;
730 731 732
}

/***********************************************************************
unknown's avatar
unknown committed
733
Frees a file node object from a tablespace memory cache. */
734 735 736 737 738
static
void
fil_node_free(
/*==========*/
	fil_node_t*	node,	/* in, own: file node */
unknown's avatar
unknown committed
739
	fil_system_t*	system,	/* in: tablespace memory cache */
740 741 742 743 744
	fil_space_t*	space)	/* in: space where the file node is chained */
{
	ut_ad(node && system && space);
	ut_ad(mutex_own(&(system->mutex)));
	ut_a(node->magic_n == FIL_NODE_MAGIC_N);
unknown's avatar
unknown committed
745
	ut_a(node->n_pending == 0);
746 747

	if (node->open) {
unknown's avatar
unknown committed
748 749 750 751 752
		/* We fool the assertion in fil_node_close_file() to think
		there are no unflushed modifications in the file */

		node->modification_counter = node->flush_counter;

unknown's avatar
unknown committed
753
		fil_node_close_file(node, system);
754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
	}

	space->size -= node->size;
	
	UT_LIST_REMOVE(chain, space->chain, node);

	mem_free(node->name);
	mem_free(node);
}

/********************************************************************
Drops files from the start of a file space, so that its size is cut by
the amount given. */

void
fil_space_truncate_start(
/*=====================*/
	ulint	id,		/* in: space id */
	ulint	trunc_len)	/* in: truncate by this much; it is an error
				if this does not equal to the combined size of
				some initial files in the space */
{
unknown's avatar
unknown committed
776
	fil_system_t*	system		= fil_system;
777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
	fil_node_t*	node;
	fil_space_t*	space;

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	ut_a(space);
	
	while (trunc_len > 0) {
		node = UT_LIST_GET_FIRST(space->chain);

		ut_a(node->size * UNIV_PAGE_SIZE >= trunc_len);

		trunc_len -= node->size * UNIV_PAGE_SIZE;

		fil_node_free(node, system, space);
	}	
				
	mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
797
}
798

unknown's avatar
unknown committed
799 800 801 802 803 804 805 806 807 808 809
/***********************************************************************
Creates a space memory object and puts it to the tablespace memory cache. If
there is an error, prints an error message to the .err log. */

ibool
fil_space_create(
/*=============*/
			/* out: TRUE if success */
	char*	name,	/* in: space name */
	ulint	id,	/* in: space id */
	ulint	purpose)/* in: FIL_TABLESPACE, or FIL_LOG if log */
810
{
unknown's avatar
unknown committed
811 812 813 814 815 816 817 818
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;	
	char*		name2;
	ulint		namesake_id;
try_again:
	/*printf(
	"InnoDB: Adding tablespace %lu of name %s, purpose %lu\n", id, name,
	  purpose);*/
819

unknown's avatar
unknown committed
820 821
	ut_a(system);
	ut_a(name);
822

unknown's avatar
unknown committed
823
	mutex_enter(&(system->mutex));
824

unknown's avatar
unknown committed
825 826 827 828 829 830 831 832 833 834
	HASH_SEARCH(name_hash, system->name_hash, ut_fold_string(name), space,
					0 == strcmp(name, space->name));
	if (space != NULL) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Warning: trying to init to the tablespace memory cache\n"
"InnoDB: a tablespace %lu of name %s,\n"
"InnoDB: but a tablespace %lu of the same name %s\n"
"InnoDB: already exists in the tablespace memory cache!\n",
			       id, name, space->id, space->name);
835

unknown's avatar
unknown committed
836
		if (id == 0 || purpose != FIL_TABLESPACE) {
837

unknown's avatar
unknown committed
838
			mutex_exit(&(system->mutex));
839

unknown's avatar
unknown committed
840 841
			return(FALSE);
		}
842

unknown's avatar
unknown committed
843 844 845 846 847 848 849 850
		fprintf(stderr,
"InnoDB: We assume that InnoDB did a crash recovery, and you had\n"
"InnoDB: an .ibd file for which the table did not exist in the\n"
"InnoDB: InnoDB internal data dictionary in the ibdata files.\n"
"InnoDB: We assume that you later removed the .ibd and .frm files,\n"
"InnoDB: and are now trying to recreate the table. We now remove the\n"
"InnoDB: conflicting tablespace object from the memory cache and try\n"
"InnoDB: the init again.\n");
851

unknown's avatar
unknown committed
852
		namesake_id = space->id;
853

unknown's avatar
unknown committed
854
		mutex_exit(&(system->mutex));
855

unknown's avatar
unknown committed
856
		fil_space_free(namesake_id);
857

unknown's avatar
unknown committed
858 859
		goto try_again;
	}
860

unknown's avatar
unknown committed
861
	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);
862

unknown's avatar
unknown committed
863 864 865 866 867 868
	if (space != NULL) {
		fprintf(stderr,
"InnoDB: Error: trying to add tablespace %lu of name %s\n"
"InnoDB: to the tablespace memory cache, but tablespace\n"
"InnoDB: %lu of name %s already exists in the tablespace\n"
"InnoDB: memory cache!\n", id, name, space->id, space->name);
869

unknown's avatar
unknown committed
870
		mutex_exit(&(system->mutex));
871

unknown's avatar
unknown committed
872
		return(FALSE);
873 874
	}

unknown's avatar
unknown committed
875
	space = mem_alloc(sizeof(fil_space_t));
876

unknown's avatar
unknown committed
877
	name2 = mem_alloc(ut_strlen(name) + 1);
878

unknown's avatar
unknown committed
879
	ut_strcpy(name2, name);
880

unknown's avatar
unknown committed
881 882
	space->name = name2;
	space->id = id;
883

unknown's avatar
unknown committed
884
	system->tablespace_version++;
unknown's avatar
unknown committed
885
	space->tablespace_version = system->tablespace_version;
unknown's avatar
unknown committed
886
	space->mark = FALSE;
887

unknown's avatar
unknown committed
888 889
	if (purpose == FIL_TABLESPACE && id > system->max_assigned_id) {
		system->max_assigned_id = id;
890 891
	}

unknown's avatar
unknown committed
892 893 894 895 896
	space->stop_ios = FALSE;
	space->stop_ibuf_merges = FALSE;
	space->is_being_deleted = FALSE;
	space->purpose = purpose;
	space->size = 0;
897

unknown's avatar
unknown committed
898
	space->n_reserved_extents = 0;
unknown's avatar
unknown committed
899
	
unknown's avatar
unknown committed
900 901
	space->n_pending_flushes = 0;
	space->n_pending_ibuf_merges = 0;
902

unknown's avatar
unknown committed
903 904
	UT_LIST_INIT(space->chain);
	space->magic_n = FIL_SPACE_MAGIC_N;
905

unknown's avatar
unknown committed
906 907 908 909 910 911
	space->ibuf_data = NULL;
	
	rw_lock_create(&(space->latch));
	rw_lock_set_level(&(space->latch), SYNC_FSP);
	
	HASH_INSERT(fil_space_t, hash, system->spaces, id, space);
912

unknown's avatar
unknown committed
913 914 915 916 917
	HASH_INSERT(fil_space_t, name_hash, system->name_hash,
						ut_fold_string(name), space);
	UT_LIST_ADD_LAST(space_list, system->space_list, space);
				
	mutex_exit(&(system->mutex));
918

unknown's avatar
unknown committed
919
	return(TRUE);
920 921 922
}

/***********************************************************************
unknown's avatar
unknown committed
923 924 925 926 927 928 929 930 931
Assigns a new space id for a new single-table tablespace. This works simply by
incrementing the global counter. If 4 billion id's is not enough, we may need
to recycle id's. */
static
ulint
fil_assign_new_space_id(void)
/*=========================*/
			/* out: new tablespace id; ULINT_UNDEFINED if could
			not assign an id */
932 933
{
	fil_system_t*	system = fil_system;
unknown's avatar
unknown committed
934
	ulint		id;
935 936 937

	mutex_enter(&(system->mutex));

unknown's avatar
unknown committed
938
	system->max_assigned_id++;
939

unknown's avatar
unknown committed
940
	id = system->max_assigned_id;
941

unknown's avatar
unknown committed
942 943 944 945 946 947 948 949 950
	if (id > (SRV_LOG_SPACE_FIRST_ID / 2) && (id % 1000000UL == 0)) {
	        ut_print_timestamp(stderr);
	        fprintf(stderr,
"InnoDB: Warning: you are running out of new single-table tablespace id's.\n"
"InnoDB: Current counter is %lu and it must not exceed %lu!\n"
"InnoDB: To reset the counter to zero you have to dump all your tables and\n"
"InnoDB: recreate the whole InnoDB installation.\n", id,
						     SRV_LOG_SPACE_FIRST_ID);
	}
951

unknown's avatar
unknown committed
952 953 954 955 956 957 958 959
	if (id >= SRV_LOG_SPACE_FIRST_ID) {
	        ut_print_timestamp(stderr);
	        fprintf(stderr,
"InnoDB: You have run out of single-table tablespace id's!\n"
"InnoDB: Current counter is %lu.\n"
"InnoDB: To reset the counter to zero you have to dump all your tables and\n"
"InnoDB: recreate the whole InnoDB installation.\n", id);
		system->max_assigned_id--;
960

unknown's avatar
unknown committed
961 962
		id = ULINT_UNDEFINED;
	}
963 964

	mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
965 966

	return(id);
967 968 969
}

/***********************************************************************
unknown's avatar
unknown committed
970 971 972
Frees a space object from the tablespace memory cache. Closes the files in
the chain but does not delete them. There must not be any pending i/o's or
flushes on the files. */
973

unknown's avatar
unknown committed
974
ibool
975 976
fil_space_free(
/*===========*/
unknown's avatar
unknown committed
977
			/* out: TRUE if success */
978 979
	ulint	id)	/* in: space id */
{
unknown's avatar
unknown committed
980
	fil_system_t*	system = fil_system;
981
	fil_space_t*	space;
unknown's avatar
unknown committed
982
	fil_space_t*	namespace;
983
	fil_node_t*	fil_node;
unknown's avatar
unknown committed
984

985 986 987 988
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
989 990 991 992 993 994 995 996 997 998 999
	if (!space) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Error: trying to remove tablespace %lu from the cache but\n"
"InnoDB: it is not there.\n", id);

		mutex_exit(&(system->mutex));
		
		return(FALSE);
	}

1000 1001
	HASH_DELETE(fil_space_t, hash, system->spaces, id, space);

unknown's avatar
unknown committed
1002 1003 1004 1005 1006 1007 1008 1009
	HASH_SEARCH(name_hash, system->name_hash, ut_fold_string(space->name),
		    namespace, 0 == strcmp(space->name, namespace->name));
	ut_a(namespace);
	ut_a(space == namespace);

	HASH_DELETE(fil_space_t, name_hash, system->name_hash,
					   ut_fold_string(space->name), space);

1010 1011 1012
	UT_LIST_REMOVE(space_list, system->space_list, space);

	ut_a(space->magic_n == FIL_SPACE_MAGIC_N);
unknown's avatar
unknown committed
1013
	ut_a(0 == space->n_pending_flushes);
1014 1015 1016 1017 1018 1019 1020 1021 1022

	fil_node = UT_LIST_GET_FIRST(space->chain);

	while (fil_node != NULL) {
		fil_node_free(fil_node, system, space);

		fil_node = UT_LIST_GET_FIRST(space->chain);
	}	
	
unknown's avatar
unknown committed
1023
	ut_a(0 == UT_LIST_GET_LEN(space->chain));
1024 1025 1026

	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
1027 1028
	rw_lock_free(&(space->latch));

1029 1030
	mem_free(space->name);
	mem_free(space);
unknown's avatar
unknown committed
1031 1032

	return(TRUE);
1033 1034
}

unknown's avatar
unknown committed
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
#ifdef UNIV_HOTBACKUP
/***********************************************************************
Returns the tablespace object for a given id, or NULL if not found from the
tablespace memory cache. */
static
fil_space_t*
fil_get_space_for_id_low(
/*=====================*/
			/* out: tablespace object or NULL; NOTE that you must
			own &(fil_system->mutex) to call this function! */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;

	ut_ad(system);

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	return(space);
}
#endif

1058
/***********************************************************************
unknown's avatar
unknown committed
1059 1060
Returns the size of the space in pages. The tablespace must be cached in the
memory cache. */
1061 1062 1063 1064

ulint
fil_space_get_size(
/*===============*/
unknown's avatar
unknown committed
1065
			/* out: space size, 0 if space not found */
1066 1067
	ulint	id)	/* in: space id */
{
unknown's avatar
unknown committed
1068 1069
	fil_system_t*	system 		= fil_system;
	fil_node_t*	node;
1070 1071 1072 1073 1074
	fil_space_t*	space;
	ulint		size;

	ut_ad(system);

unknown's avatar
unknown committed
1075
	fil_mutex_enter_and_prepare_for_io(id);
1076 1077 1078

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
	if (space == NULL) {
		mutex_exit(&(system->mutex));

		return(0);
	}

	if (space->size == 0 && space->purpose == FIL_TABLESPACE) {
		ut_a(id != 0);

		ut_a(1 == UT_LIST_GET_LEN(space->chain));

		node = UT_LIST_GET_FIRST(space->chain);

		/* It must be a single-table tablespace and we have not opened
		the file yet; the following calls will open it and update the
		size fields */

		fil_node_prepare_for_io(node, system, space);
		fil_node_complete_io(node, system, OS_FILE_READ);
	}

1100 1101 1102 1103
	size = space->size;
	
	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354
	return(size);
}

/***********************************************************************
Checks if the pair space, page_no refers to an existing page in a tablespace
file space. The tablespace must be cached in the memory cache. */

ibool
fil_check_adress_in_tablespace(
/*===========================*/
			/* out: TRUE if the address is meaningful */
	ulint	id,	/* in: space id */
	ulint	page_no)/* in: page number */
{
	if (fil_space_get_size(id) > page_no) {

		return(TRUE);
	}

	return(FALSE);
}		

/********************************************************************
Creates a the tablespace memory cache. */
static
fil_system_t*
fil_system_create(
/*==============*/
				/* out, own: tablespace memory cache */
	ulint	hash_size,	/* in: hash table size */
	ulint	max_n_open)	/* in: maximum number of open files; must be
				> 10 */
{
	fil_system_t*	system;

	ut_a(hash_size > 0);
	ut_a(max_n_open > 0);

	system = mem_alloc(sizeof(fil_system_t));

	mutex_create(&(system->mutex));

	mutex_set_level(&(system->mutex), SYNC_ANY_LATCH);

	system->spaces = hash_create(hash_size);
	system->name_hash = hash_create(hash_size);

	UT_LIST_INIT(system->LRU);

	system->n_open = 0;
	system->max_n_open = max_n_open;

	system->modification_counter = 0;
	system->max_assigned_id = 0;

	system->tablespace_version = 0;

	UT_LIST_INIT(system->space_list);

	return(system);
}

/********************************************************************
Initializes the tablespace memory cache. */

void
fil_init(
/*=====*/
	ulint	max_n_open)	/* in: max number of open files */
{
	ut_a(fil_system == NULL);

	/*printf("Initializing the tablespace cache with max %lu open files\n",
							       max_n_open); */
	fil_system = fil_system_create(FIL_SYSTEM_HASH_SIZE, max_n_open);
}

/***********************************************************************
Opens all log files and system tablespace data files. They stay open until the
database server shutdown. This should be called at a server startup after the
space objects for the log and the system tablespace have been created. The
purpose of this operation is to make sure we never run out of file descriptors
if we need to read from the insert buffer or to write to the log. */

void
fil_open_log_and_system_tablespace_files(void)
/*==========================================*/
{
	fil_system_t*	system = fil_system;
	fil_space_t*	space;
	fil_node_t*	node;

	mutex_enter(&(system->mutex));

	space = UT_LIST_GET_FIRST(system->space_list);

	while (space != NULL) {
		if (space->purpose != FIL_TABLESPACE || space->id == 0) {
			node = UT_LIST_GET_FIRST(space->chain);

			while (node != NULL) {
				if (!node->open) {
					fil_node_open_file(node, system,
									space);
				}
				if (system->max_n_open < 10 + system->n_open) {
					fprintf(stderr,
"InnoDB: Warning: you must raise the value of innodb_max_open_files in\n"
"InnoDB: my.cnf! Remember that InnoDB keeps all log files and all system\n"
"InnoDB: tablespace files open for the whole time mysqld is running, and\n"
"InnoDB: needs to open also some .ibd files if the file-per-table storage\n"
"InnoDB: model is used. Current open files %lu, max allowed open files %lu.\n",
				     system->n_open, system->max_n_open);
				}
				node = UT_LIST_GET_NEXT(chain, node);
			}
		}
		space = UT_LIST_GET_NEXT(space_list, space);
	}

	mutex_exit(&(system->mutex));
}

/***********************************************************************
Closes all open files. There must not be any pending i/o's or not flushed
modifications in the files. */

void
fil_close_all_files(void)
/*=====================*/
{
	fil_system_t*	system = fil_system;
	fil_space_t*	space;
	fil_node_t*	node;

	mutex_enter(&(system->mutex));

	space = UT_LIST_GET_FIRST(system->space_list);

	while (space != NULL) {
		node = UT_LIST_GET_FIRST(space->chain);

		while (node != NULL) {
			if (node->open) {
				fil_node_close_file(node, system);
			}
			node = UT_LIST_GET_NEXT(chain, node);
		}
		space = UT_LIST_GET_NEXT(space_list, space);
	}

	mutex_exit(&(system->mutex));
}

/***********************************************************************
Sets the max tablespace id counter if the given number is bigger than the
previous value. */

void
fil_set_max_space_id_if_bigger(
/*===========================*/
	ulint	max_id)	/* in: maximum known id */
{
	fil_system_t*	system = fil_system;

	if (max_id >= SRV_LOG_SPACE_FIRST_ID) {
		fprintf(stderr,
"InnoDB: Fatal error: max tablespace id is too high, %lu\n", max_id);
		ut_a(0);
	}

	mutex_enter(&(system->mutex));

	if (system->max_assigned_id < max_id) {

		system->max_assigned_id = max_id;
	}

	mutex_exit(&(system->mutex));
}

/********************************************************************
Initializes the ibuf data structure for space 0 == the system tablespace.
This can be called after the file space headers have been created and the
dictionary system has been initialized. */

void
fil_ibuf_init_at_db_start(void)
/*===========================*/
{
	fil_space_t*	space;

	space = UT_LIST_GET_FIRST(fil_system->space_list);

	ut_a(space);
        ut_a(space->purpose == FIL_TABLESPACE);	

	space->ibuf_data = ibuf_data_init_for_space(space->id);
}

/********************************************************************
Writes the flushed lsn and the latest archived log number to the page header
of the first page of a data file. */
static
ulint
fil_write_lsn_and_arch_no_to_file(
/*==============================*/
	ulint	space_id,	/* in: space number */
	ulint	sum_of_sizes,	/* in: combined size of previous files in
				space, in database pages */
	dulint	lsn,		/* in: lsn to write */
	ulint	arch_log_no)	/* in: archived log number to write */
{
	byte*	buf1;
	byte*	buf;

	UT_NOT_USED(arch_log_no);

	buf1 = mem_alloc(2 * UNIV_PAGE_SIZE);
	buf = ut_align(buf1, UNIV_PAGE_SIZE);

	fil_read(TRUE, space_id, sum_of_sizes, 0, UNIV_PAGE_SIZE, buf, NULL);

	mach_write_to_8(buf + FIL_PAGE_FILE_FLUSH_LSN, lsn);

	fil_write(TRUE, space_id, sum_of_sizes, 0, UNIV_PAGE_SIZE, buf, NULL);

	return(DB_SUCCESS);	
}

/********************************************************************
Writes the flushed lsn and the latest archived log number to the page
header of the first page of each data file in the system tablespace. */

ulint
fil_write_flushed_lsn_to_data_files(
/*================================*/
				/* out: DB_SUCCESS or error number */
	dulint	lsn,		/* in: lsn to write */
	ulint	arch_log_no)	/* in: latest archived log file number */
{
	fil_space_t*	space;
	fil_node_t*	node;
	ulint		sum_of_sizes;
	ulint		err;

	mutex_enter(&(fil_system->mutex));
	
	space = UT_LIST_GET_FIRST(fil_system->space_list);
	
	while (space) {
1355 1356 1357 1358 1359
		/* We only write the lsn to all existing data files which have
		been open during the lifetime of the mysqld process; they are
		represented by the space objects in the tablespace memory
		cache. Note that all data files in the system tablespace 0 are
		always open. */
unknown's avatar
unknown committed
1360

1361
		if (space->purpose == FIL_TABLESPACE) {
unknown's avatar
unknown committed
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508
			sum_of_sizes = 0;

			node = UT_LIST_GET_FIRST(space->chain);
			while (node) {
				mutex_exit(&(fil_system->mutex));

				err = fil_write_lsn_and_arch_no_to_file(
						space->id, sum_of_sizes,
						lsn, arch_log_no);
				if (err != DB_SUCCESS) {

					return(err);
				}

				mutex_enter(&(fil_system->mutex));

				sum_of_sizes += node->size;
				node = UT_LIST_GET_NEXT(chain, node);
			}
		}
		space = UT_LIST_GET_NEXT(space_list, space);
	}

	mutex_exit(&(fil_system->mutex));

	return(DB_SUCCESS);
}

/***********************************************************************
Reads the flushed lsn and arch no fields from a data file at database
startup. */

void
fil_read_flushed_lsn_and_arch_log_no(
/*=================================*/
	os_file_t data_file,		/* in: open data file */
	ibool	one_read_already,	/* in: TRUE if min and max parameters
					below already contain sensible data */
	dulint*	min_flushed_lsn,	/* in/out: */
	ulint*	min_arch_log_no,	/* in/out: */
	dulint*	max_flushed_lsn,	/* in/out: */
	ulint*	max_arch_log_no)	/* in/out: */
{
	byte*	buf;
	byte*	buf2;
	dulint	flushed_lsn;
        ulint   arch_log_no     = 0;    /* since InnoDB does not archive
                                        its own logs under MySQL, this
                                        parameter is not relevant */
	buf2 = ut_malloc(2 * UNIV_PAGE_SIZE);
	/* Align the memory for a possible read from a raw device */
	buf = ut_align(buf2, UNIV_PAGE_SIZE);
	
	os_file_read(data_file, buf, 0, 0, UNIV_PAGE_SIZE);

	flushed_lsn = mach_read_from_8(buf + FIL_PAGE_FILE_FLUSH_LSN);

	ut_free(buf2);

	if (!one_read_already) {
		*min_flushed_lsn = flushed_lsn;
		*max_flushed_lsn = flushed_lsn;
		*min_arch_log_no = arch_log_no;
		*max_arch_log_no = arch_log_no;

		return;
	}

	if (ut_dulint_cmp(*min_flushed_lsn, flushed_lsn) > 0) {
		*min_flushed_lsn = flushed_lsn;
	}
	if (ut_dulint_cmp(*max_flushed_lsn, flushed_lsn) < 0) {
		*max_flushed_lsn = flushed_lsn;
	}
	if (*min_arch_log_no > arch_log_no) {
		*min_arch_log_no = arch_log_no;
	}
	if (*max_arch_log_no < arch_log_no) {
		*max_arch_log_no = arch_log_no;
	}
}

/*================ SINGLE-TABLE TABLESPACES ==========================*/

/***********************************************************************
Increments the count of pending insert buffer page merges, if space is not
being deleted. */

ibool
fil_inc_pending_ibuf_merges(
/*========================*/
			/* out: TRUE if being deleted, and ibuf merges should
			be skipped */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;
	
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL) {
		fprintf(stderr,
"InnoDB: Error: trying to do ibuf merge to a dropped tablespace %lu\n", id);
	}

	if (space == NULL || space->stop_ibuf_merges) {
		mutex_exit(&(system->mutex));

		return(TRUE);
	}

	space->n_pending_ibuf_merges++;

	mutex_exit(&(system->mutex));

	return(FALSE);
}

/***********************************************************************
Decrements the count of pending insert buffer page merges. */

void
fil_decr_pending_ibuf_merges(
/*========================*/
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;
	
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL) {
		fprintf(stderr,
"InnoDB: Error: decrementing ibuf merge of a dropped tablespace %lu\n", id);
	}

	if (space != NULL) {
		space->n_pending_ibuf_merges--;
	}

	mutex_exit(&(system->mutex));
}

unknown's avatar
unknown committed
1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727
/************************************************************
Creates the database directory for a table if it does not exist yet. */
static
void
fil_create_directory_for_tablename(
/*===============================*/
	char*	name)	/* in: name in the standard 'databasename/tablename'
			format */
{
	char*	ptr;
	char	path[OS_FILE_MAX_PATH];

	sprintf(path, "%s/%s", fil_path_to_mysql_datadir, name);

	ptr = path + ut_strlen(path);

	while (*ptr != '/') {
		ptr--;

		ut_a(ptr >= path);
	}

	*ptr = '\0';

	srv_normalize_path_for_win(path);

	ut_a(os_file_create_directory(path, FALSE));
}

#ifndef UNIV_HOTBACKUP
/************************************************************
Writes a log record about an .ibd file create/rename/delete. */
static
void
fil_op_write_log(
/*=============*/
	ulint	type,		/* in: MLOG_FILE_CREATE, MLOG_FILE_DELETE, or
                        	MLOG_FILE_RENAME */
	ulint	space_id,	/* in: space id */
	char*	name,		/* in: table name in the familiar
				'databasename/tablename' format, or the file
				path in the case of MLOG_FILE_DELETE */ 
	char*	new_name,	/* in: if type is MLOG_FILE_RENAME, the new
				table name in the 'databasename/tablename'
				format */
	mtr_t*	mtr)		/* in: mini-transaction handle */
{
	byte*	log_ptr;

	log_ptr = mlog_open(mtr, 30);
	
	log_ptr = mlog_write_initial_log_record_for_file_op(type, space_id, 0,
								log_ptr, mtr);
	/* Let us store the strings as null-terminated for easier readability
	and handling */

	mach_write_to_2(log_ptr, ut_strlen(name) + 1);
	log_ptr += 2;
	
	mlog_close(mtr, log_ptr);

	mlog_catenate_string(mtr, name, ut_strlen(name) + 1);

	if (type == MLOG_FILE_RENAME) {
		log_ptr = mlog_open(mtr, 30);
		mach_write_to_2(log_ptr, ut_strlen(new_name) + 1);
		log_ptr += 2;
	
		mlog_close(mtr, log_ptr);

		mlog_catenate_string(mtr, new_name, ut_strlen(new_name) + 1);
	}
}
#endif

/***********************************************************************
Parses the body of a log record written about an .ibd file operation. That is,
the log record part after the standard (type, space id, page no) header of the
log record.

If desired, also replays the delete or rename operation if the .ibd file
exists and the space id in it matches. Replays the create operation if a file
at that path does not exist yet. If the database directory for the file to be
created does not exist, then we create the directory, too.

Note that ibbackup --apply-log sets fil_path_to_mysql_datadir to point to the
datadir that we should use in replaying the file operations. */

byte*
fil_op_log_parse_or_replay(
/*=======================*/
                        	/* out: end of log record, or NULL if the
				record was not completely contained between
				ptr and end_ptr */
        byte*   ptr,    	/* in: buffer containing the log record body,
				or an initial segment of it, if the record does
				not fir completely between ptr and end_ptr */
        byte*   end_ptr,	/* in: buffer end */
	ulint	type,		/* in: the type of this log record */
	ibool	do_replay,	/* in: TRUE if we want to replay the
				operation, and not just parse the log record */
	ulint	space_id)	/* in: if do_replay is TRUE, the space id of
				the tablespace in question; otherwise
				ignored */
{
	ulint	name_len;
	ulint	new_name_len;
	char*	name;
	char*	new_name	= NULL;

	if (end_ptr < ptr + 2) {

		return(NULL);
	}

	name_len = mach_read_from_2(ptr);

	ptr += 2;

	if (end_ptr < ptr + name_len) {
		
		return(NULL);
	}

	name = ptr;

	ptr += name_len;

	if (type == MLOG_FILE_RENAME) {
		if (end_ptr < ptr + 2) {

			return(NULL);
		}

		new_name_len = mach_read_from_2(ptr);
		
		ptr += 2;

		if (end_ptr < ptr + new_name_len) {
		
			return(NULL);
		}

		new_name = ptr;

		ptr += new_name_len;
	}

	/* We managed to parse a full log record body */
/*
	printf("Parsed log rec of type %lu space %lu\n"
		"name %s\n", type, space_id, name);

	if (type == MLOG_FILE_RENAME) {
		printf("new name %s\n", new_name);
	}
*/
	if (do_replay == FALSE) {

		return(ptr);
	}

	/* Let us try to perform the file operation, if sensible. Note that
	ibbackup has at this stage already read in all space id info to the
	fil0fil.c data structures.
	
	NOTE that our algorithm is not guaranteed to work correctly if there
	were renames of tables during the backup. See ibbackup code for more
	on the problem. */

	if (type == MLOG_FILE_DELETE) {
		if (fil_tablespace_exists_in_mem(space_id)) {
			ut_a(fil_delete_tablespace(space_id));
		}
	} else if (type == MLOG_FILE_RENAME) {
		if (fil_get_space_id_for_table(name) == space_id) {
			/* Create the database directory for the new name, if
			it does not exist yet */
			fil_create_directory_for_tablename(new_name);
	
			/* Rename the table if there is not yet a tablespace
			with the same name */

			if (fil_get_space_id_for_table(new_name)
			    == ULINT_UNDEFINED) {
				ut_a(fil_rename_tablespace(name, space_id,
								new_name));
			} else {
				fprintf(stderr,
"InnoDB: Warning: in log replay cannot rename tablespace\n"
"InnoDB: %s with id %lu to %s, because it exists already.\n",
					name, space_id, new_name);
			}
		}
	} else {
		ut_a(type == MLOG_FILE_CREATE);

		if (fil_tablespace_exists_in_mem(space_id)) {
			/* Do nothing */
		} else if (fil_get_space_id_for_table(name) !=
							ULINT_UNDEFINED) {
			/* Do nothing */
		} else {
			/* Create the database directory for name, if it does
			not exist yet */
			fil_create_directory_for_tablename(name);

			ut_a(space_id != 0);

			ut_a(DB_SUCCESS == 
				fil_create_new_single_table_tablespace(
							&space_id, name,
						FIL_IBD_FILE_INITIAL_SIZE));
		}
	}

	return(ptr);
}

unknown's avatar
unknown committed
1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788
/***********************************************************************
Deletes a single-table tablespace. The tablespace must be cached in the
memory cache. */

ibool
fil_delete_tablespace(
/*==================*/
			/* out: TRUE if success */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	ibool		success;
	fil_space_t*	space;
	fil_node_t*	node;
	ulint		count		= 0;
	char		path[OS_FILE_MAX_PATH];

	ut_a(id != 0);
stop_ibuf_merges:
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space != NULL) {
		space->stop_ibuf_merges = TRUE;

		if (space->n_pending_ibuf_merges == 0) {
			mutex_exit(&(system->mutex));

			count = 0;

			goto try_again;
		} else {
			if (count > 5000) {
			   ut_print_timestamp(stderr);
			   fprintf(stderr,
"  InnoDB: Warning: trying to delete tablespace %s,\n"
"InnoDB: but there are %lu pending ibuf merges on it.\n"
"InnoDB: Loop %lu.\n", space->name, space->n_pending_ibuf_merges, count);
			}

			mutex_exit(&(system->mutex));

			os_thread_sleep(20000);
			count++;

			goto stop_ibuf_merges;
		}
	}

	mutex_exit(&(system->mutex));
	count = 0;

try_again:
	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
unknown's avatar
unknown committed
1789 1790
"  InnoDB: Error: cannot delete tablespace %lu\n"
"InnoDB: because it is not found in the tablespace memory cache.\n", id);
unknown's avatar
unknown committed
1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825

		mutex_exit(&(system->mutex));
	
		return(FALSE);
	}	

	ut_a(space);
	ut_a(strlen(space->name) < OS_FILE_MAX_PATH);
	ut_a(space->n_pending_ibuf_merges == 0);

	strcpy(path, space->name);

	space->is_being_deleted = TRUE;

	ut_a(UT_LIST_GET_LEN(space->chain) == 1);
	node = UT_LIST_GET_FIRST(space->chain);

	if (space->n_pending_flushes > 0 || node->n_pending > 0) {
		if (count > 1000) {
			ut_print_timestamp(stderr);
			fprintf(stderr,
"  InnoDB: Warning: trying to delete tablespace %s,\n"
"InnoDB: but there are %lu flushes and %lu pending i/o's on it\n"
"InnoDB: Loop %lu.\n", space->name, space->n_pending_flushes, node->n_pending,
									count);
		}
		mutex_exit(&(system->mutex));
		os_thread_sleep(20000);

		count++;

		goto try_again;
	}

	mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
1826
#ifndef UNIV_HOTBACKUP
unknown's avatar
unknown committed
1827 1828 1829 1830 1831 1832 1833 1834
	/* Invalidate in the buffer pool all pages belonging to the
	tablespace. Since we have set space->is_being_deleted = TRUE, readahead
	or ibuf merge can no longer read more pages of this tablespace to the
	buffer pool. Thus we can clean the tablespace out of the buffer pool
	completely and permanently. The flag is_being_deleted also prevents
	fil_flush() from being applied to this tablespace. */

	buf_LRU_invalidate_tablespace(id);
unknown's avatar
unknown committed
1835 1836
#endif
	/* printf("Deleting tablespace %s id %lu\n", space->name, id); */
unknown's avatar
unknown committed
1837 1838 1839 1840 1841 1842 1843

	success = fil_space_free(id);

	if (success) {
		success = os_file_delete(path);

		if (success) {
unknown's avatar
unknown committed
1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860
			/* Write a log record about the deletion of the .ibd
			file, so that ibbackup can replay it in the
			--apply-log phase. We use a dummy mtr and the familiar
			log write mechanism. */
#ifndef UNIV_HOTBACKUP
			{
			mtr_t		mtr;

			/* When replaying the operation in ibbackup, do not try
			to write any log record */
			mtr_start(&mtr);

			fil_op_write_log(MLOG_FILE_DELETE, id, path,
								NULL, &mtr);
			mtr_commit(&mtr);
			}
#endif
unknown's avatar
unknown committed
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968
			return(TRUE);
		}
	}

	return(FALSE);
}

/***********************************************************************
Discards a single-table tablespace. The tablespace must be cached in the
memory cache. Discarding is like deleting a tablespace, but
1) we do not drop the table from the data dictionary;
2) we remove all insert buffer entries for the tablespace immediately; in DROP
TABLE they are only removed gradually in the background;
3) when the user does IMPORT TABLESPACE, the tablespace will have the same id
as it originally had. */

ibool
fil_discard_tablespace(
/*===================*/
			/* out: TRUE if success */
	ulint	id)	/* in: space id */
{
	ibool	success;

	success = fil_delete_tablespace(id);

	if (!success) {
		fprintf(stderr,
"InnoDB: Warning: cannot delete tablespace %lu in DISCARD TABLESPACE.\n"
"InnoDB: But let us remove the insert buffer entries for this tablespace.\n",
									id); 
	}

	/* Remove all insert buffer entries for the tablespace */

	ibuf_delete_for_discarded_space(id);

	return(TRUE);
}

/***********************************************************************
Renames the memory cache structures of a single-table tablespace. */
static
ibool
fil_rename_tablespace_in_mem(
/*=========================*/
				/* out: TRUE if success */
	fil_space_t*	space,	/* in: tablespace memory object */
	fil_node_t*	node,	/* in: file node of that tablespace */
	char*		path)	/* in: new name */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space2;
	char*		old_name	= space->name;
	
	HASH_SEARCH(name_hash, system->name_hash, ut_fold_string(old_name),
			       space2, 0 == strcmp(old_name, space2->name));
	if (space != space2) {
		fprintf(stderr,
"InnoDB: Error: cannot find %s in tablespace memory cache\n", old_name);

		return(FALSE);
	}

	HASH_SEARCH(name_hash, system->name_hash, ut_fold_string(path),
			       space2, 0 == strcmp(path, space2->name));
	if (space2 != NULL) {
		fprintf(stderr,
"InnoDB: Error: %s is already in tablespace memory cache\n", path);
		
		return(FALSE);
	}

	HASH_DELETE(fil_space_t, name_hash, system->name_hash,
					   ut_fold_string(space->name), space);
	mem_free(space->name);
	mem_free(node->name);

	space->name = mem_alloc(strlen(path) + 1);
	node->name = mem_alloc(strlen(path) + 1);

	strcpy(space->name, path);
	strcpy(node->name, path);

	HASH_INSERT(fil_space_t, name_hash, system->name_hash,
						ut_fold_string(path), space);
	return(TRUE);
}

/***********************************************************************
Renames a single-table tablespace. The tablespace must be cached in the
tablespace memory cache. */

ibool
fil_rename_tablespace(
/*==================*/
				/* out: TRUE if success */
	char*	old_name,	/* in: old table name in the standard
				databasename/tablename format of InnoDB */
	ulint	id,		/* in: space id */
	char*	new_name)	/* in: new table name in the standard
				databasename/tablename format of InnoDB */
{
	fil_system_t*	system		= fil_system;
	ibool		success;
	fil_space_t*	space;
	fil_node_t*	node;
	ulint		count		= 0;
unknown's avatar
unknown committed
1969
	char*		path		= NULL;
unknown's avatar
unknown committed
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041
	char		old_path[OS_FILE_MAX_PATH];

	ut_a(id != 0);
retry:
	count++;

	if (count > 1000) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Warning: problems renaming %s to %s, %lu iterations\n",
						old_name, new_name, count);
	}

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL) {
		fprintf(stderr,
"InnoDB: Error: cannot find space id %lu from the tablespace memory cache\n"
"InnoDB: though the table %s in a rename operation should have that id\n",
							  id, old_name);
		mutex_exit(&(system->mutex));

		return(FALSE);
	}

	if (count > 25000) {
		space->stop_ios = FALSE;
		mutex_exit(&(system->mutex));

		return(FALSE);
	}

	/* We temporarily close the .ibd file because we do not trust that
	operating systems can rename an open file. For the closing we have to
	wait until there are no pending i/o's or flushes on the file. */

	space->stop_ios = TRUE;

	ut_a(UT_LIST_GET_LEN(space->chain) == 1);
	node = UT_LIST_GET_FIRST(space->chain);

	if (node->n_pending > 0 || node->n_pending_flushes > 0) {
		/* There are pending i/o's or flushes, sleep for a while and
		retry */

		mutex_exit(&(system->mutex));

		os_thread_sleep(20000);

		goto retry;

	} else if (node->modification_counter > node->flush_counter) {
		/* Flush the space */

		mutex_exit(&(system->mutex));

		os_thread_sleep(20000);

		fil_flush(id);

		goto retry;

	} else if (node->open) {
		/* Close the file */

		fil_node_close_file(node, system);
	}

	/* Check that the old name in the space is right */
	
unknown's avatar
unknown committed
2042 2043
	ut_a(strlen(old_name) + strlen(fil_path_to_mysql_datadir)
						< OS_FILE_MAX_PATH - 10);
unknown's avatar
unknown committed
2044

unknown's avatar
unknown committed
2045
	sprintf(old_path, "%s/%s.ibd", fil_path_to_mysql_datadir, old_name);
unknown's avatar
unknown committed
2046 2047 2048 2049 2050 2051 2052 2053

	srv_normalize_path_for_win(old_path);

	ut_a(strcmp(space->name, old_path) == 0);
	ut_a(strcmp(node->name, old_path) == 0);

	/* Rename the tablespace and the node in the memory cache */
	
unknown's avatar
unknown committed
2054 2055 2056
	ut_a(strlen(new_name) + strlen(fil_path_to_mysql_datadir)
						< OS_FILE_MAX_PATH - 10);
	path = mem_alloc(OS_FILE_MAX_PATH);
unknown's avatar
unknown committed
2057

unknown's avatar
unknown committed
2058
	sprintf(path, "%s/%s.ibd", fil_path_to_mysql_datadir, new_name);
unknown's avatar
unknown committed
2059 2060 2061 2062 2063 2064 2065 2066 2067 2068

	srv_normalize_path_for_win(path);

	success = fil_rename_tablespace_in_mem(space, node, path);

	if (!success) {

		goto func_exit;	
	}

unknown's avatar
unknown committed
2069 2070
	/* printf("Renaming tablespace %s to %s id %lu\n", path, old_path, id);
	*/
unknown's avatar
unknown committed
2071 2072 2073 2074 2075 2076 2077 2078
	success = os_file_rename(old_path, path);

	if (!success) {
		/* We have to revert the changes we made to the tablespace
		memory cache */

		ut_a(fil_rename_tablespace_in_mem(space, node, old_path));
	}
unknown's avatar
unknown committed
2079

unknown's avatar
unknown committed
2080
func_exit:
unknown's avatar
unknown committed
2081 2082 2083
	if (path) {
		mem_free(path);
	}
unknown's avatar
unknown committed
2084 2085 2086 2087
	space->stop_ios = FALSE;

	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098
#ifndef UNIV_HOTBACKUP	
	if (success) {
		mtr_t		mtr;

		mtr_start(&mtr);

		fil_op_write_log(MLOG_FILE_RENAME, id, old_name, new_name,
								&mtr);
		mtr_commit(&mtr);
	}
#endif
unknown's avatar
unknown committed
2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
	return(success);
}

/***********************************************************************
Creates a new single-table tablespace to a database directory of MySQL.
Database directories are under the 'datadir' of MySQL. The datadir is the
directory of a running mysqld program. We can refer to it by simply the
path '.'. */

ulint
fil_create_new_single_table_tablespace(
/*===================================*/
				/* out: DB_SUCCESS or error code */
unknown's avatar
unknown committed
2112 2113 2114
	ulint*	space_id,	/* in/out: space id; if this is != 0, then
				this is an input parameter, otherwise
				output */
unknown's avatar
unknown committed
2115 2116 2117
	char*	tablename,	/* in: the table name in the usual
				databasename/tablename format of InnoDB */
	ulint	size)		/* in: the initial size of the tablespace file
unknown's avatar
unknown committed
2118 2119
				in pages, must be >= FIL_IBD_FILE_INITIAL_SIZE
				*/
unknown's avatar
unknown committed
2120 2121 2122 2123 2124 2125 2126 2127
{
	os_file_t       file;
	ibool		ret;
	ulint		err;
	byte*		page;
	ibool		success;
	char		path[OS_FILE_MAX_PATH];

unknown's avatar
unknown committed
2128
	ut_a(size >= FIL_IBD_FILE_INITIAL_SIZE);
unknown's avatar
unknown committed
2129

unknown's avatar
unknown committed
2130 2131 2132
	ut_a(strlen(tablename) + strlen(fil_path_to_mysql_datadir)
						< OS_FILE_MAX_PATH - 10);
	sprintf(path, "%s/%s.ibd", fil_path_to_mysql_datadir, tablename);
unknown's avatar
unknown committed
2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179

	srv_normalize_path_for_win(path);
	
	file = os_file_create(path, OS_FILE_CREATE, OS_FILE_NORMAL,
						    OS_DATA_FILE, &ret);
	if (ret == FALSE) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Error creating file %s.\n", path);

		/* The following call will print an error message */
		 
		err = os_file_get_last_error(TRUE);
		
		if (err == OS_FILE_ALREADY_EXISTS) {
		        fprintf(stderr,
"InnoDB: The file already exists though the corresponding table did not\n"
"InnoDB: exist in the InnoDB data dictionary. Have you moved InnoDB\n"
"InnoDB: .ibd files around without using the SQL commands\n"
"InnoDB: DISCARD TABLESPACE and IMPORT TABLESPACE, or did\n"
"InnoDB: mysqld crash in the middle of CREATE TABLE? You can\n"
"InnoDB: resolve the problem by removing the file %s\n"
"InnoDB: under the 'datadir' of MySQL.\n", path);

			return(DB_TABLESPACE_ALREADY_EXISTS);
		}

		if (err == OS_FILE_DISK_FULL) {

			return(DB_OUT_OF_FILE_SPACE);
		}

		return(DB_ERROR);
	}

	page = ut_malloc(UNIV_PAGE_SIZE);

	ret = os_file_set_size(path, file, size * UNIV_PAGE_SIZE, 0);
	
	if (!ret) {
		ut_free(page);
		os_file_close(file);
		os_file_delete(path);

		return(DB_OUT_OF_FILE_SPACE);
	}

unknown's avatar
unknown committed
2180 2181 2182 2183 2184
	if (*space_id == 0) {
		*space_id = fil_assign_new_space_id();
	}

	/* printf("Creating tablespace %s id %lu\n", path, *space_id); */
unknown's avatar
unknown committed
2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252

	if (*space_id == ULINT_UNDEFINED) {
		ut_free(page);
		os_file_close(file);
		os_file_delete(path);

		return(DB_ERROR);
	}

	/* We have to write the space id to the file immediately and flush the
	file to disk. This is because in crash recovery we must be aware what
	tablespaces exist and what are their space id's, so that we can apply
	the log records to the right file. It may take quite a while until
	buffer pool flush algorithms write anything to the file and flush it to
	disk. If we would not write here anything, the file would be filled
	with zeros from the call of os_file_set_size(), until a buffer pool
	flush would write to it. */

	memset(page, '\0', UNIV_PAGE_SIZE);

	fsp_header_write_space_id(page, *space_id);		

	buf_flush_init_for_writing(page, ut_dulint_zero, *space_id, 0);

	ret = os_file_write(path, file, page, 0, 0, UNIV_PAGE_SIZE);

	ut_free(page);

	if (!ret) {
		fprintf(stderr,
"InnoDB: Error: could not write the first page to tablespace %s\n", path);

		os_file_close(file);
		os_file_delete(path);

		return(DB_ERROR);
	}

	ret = os_file_flush(file);

	if (!ret) {
		fprintf(stderr,
"InnoDB: Error: file flush of tablespace %s failed\n", path);

		os_file_close(file);
		os_file_delete(path);

		return(DB_ERROR);
	}

	os_file_close(file);

	if (*space_id == ULINT_UNDEFINED) {
		os_file_delete(path);

		return(DB_ERROR);
	}

	success = fil_space_create(path, *space_id, FIL_TABLESPACE);
	
	if (!success) {
		os_file_delete(path);

		return(DB_ERROR);
	}	

	fil_node_create(path, size, *space_id, FALSE);

unknown's avatar
unknown committed
2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263
#ifndef UNIV_HOTBACKUP	
	{
	mtr_t		mtr;

	mtr_start(&mtr);

	fil_op_write_log(MLOG_FILE_CREATE, *space_id, tablename, NULL, &mtr);

	mtr_commit(&mtr);
	}
#endif
unknown's avatar
unknown committed
2264 2265 2266
	return(DB_SUCCESS);
}

2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300
/************************************************************************
It is possible, though very improbable, that the lsn's in the tablespace to be
imported have risen above the current system lsn, if a lengthy purge, ibuf
merge, or rollback was performed on a backup taken with ibbackup. If that is
the case, reset page lsn's in the file. We assume that mysqld was shut down
after it performed these cleanup operations on the .ibd file, so that it at
the shutdown stamped the latest lsn to the FIL_PAGE_FILE_FLUSH_LSN in the
first page of the .ibd file, and we can determine whether we need to reset the
lsn's just by looking at that flush lsn. */

ibool
fil_reset_too_high_lsns(
/*====================*/
				/* out: TRUE if success */
	char*	name,		/* in: table name in the databasename/tablename
				format */
	dulint	current_lsn)	/* in: reset lsn's if the lsn stamped to
				FIL_PAGE_FILE_FLUSH_LSN in the first page is
				too high */
{
	os_file_t	file;
	char*		filepath;
	byte*		page;
	dulint		flush_lsn;
	ulint		space_id;
	ib_longlong	file_size;
	ib_longlong	offset;
	ulint		page_no;
	ibool		success;

	filepath = ut_malloc(OS_FILE_MAX_PATH);

	ut_a(strlen(name) < OS_FILE_MAX_PATH - 10);

unknown's avatar
unknown committed
2301
	sprintf(filepath, "%s/%s.ibd", fil_path_to_mysql_datadir, name);
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407
					
	srv_normalize_path_for_win(filepath);

	file = os_file_create_simple_no_error_handling(filepath, OS_FILE_OPEN,
						OS_FILE_READ_WRITE, &success);
	if (!success) {
		ut_free(filepath);

		return(FALSE);
	}

	/* Read the first page of the tablespace */

	page = ut_malloc(UNIV_PAGE_SIZE);

	success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
	if (!success) {

		goto func_exit;
	}

	/* We have to read the file flush lsn from the header of the file */

	flush_lsn = mach_read_from_8(page + FIL_PAGE_FILE_FLUSH_LSN);

	if (ut_dulint_cmp(current_lsn, flush_lsn) >= 0) {
		/* Ok */
		success = TRUE;

		goto func_exit;
	}

	space_id = fsp_header_get_space_id(page);
	
	ut_print_timestamp(stderr);
	fprintf(stderr,
" InnoDB: Flush lsn in the tablespace file %lu to be imported\n"
"InnoDB: is %lu %lu, which exceeds current system lsn %lu %lu.\n"
"InnoDB: We reset the lsn's in the file %s.\n",
			    space_id,
			    ut_dulint_get_high(flush_lsn),
			    ut_dulint_get_low(flush_lsn),
			    ut_dulint_get_high(current_lsn),
			    ut_dulint_get_low(current_lsn), filepath);

	/* Loop through all the pages in the tablespace and reset the lsn and
	the page checksum if necessary */

	file_size = os_file_get_size_as_iblonglong(file);

	for (offset = 0; offset < file_size; offset += UNIV_PAGE_SIZE) {
		success = os_file_read(file, page,
				(ulint)(offset & 0xFFFFFFFFUL),
				(ulint)(offset >> 32), UNIV_PAGE_SIZE);
		if (!success) {

			goto func_exit;
		}
		if (ut_dulint_cmp(mach_read_from_8(page + FIL_PAGE_LSN),
				  current_lsn) > 0) {
			/* We have to reset the lsn */
			space_id = mach_read_from_4(page
					+ FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID);
			page_no = mach_read_from_4(page + FIL_PAGE_OFFSET);
			
			buf_flush_init_for_writing(page, current_lsn, space_id,
								      page_no);
			success = os_file_write(filepath, file, page,
				(ulint)(offset & 0xFFFFFFFFUL),
				(ulint)(offset >> 32), UNIV_PAGE_SIZE);
			if (!success) {

				goto func_exit;
			}
		}
	}

	success = os_file_flush(file);
	if (!success) {

		goto func_exit;
	}

	/* We now update the flush_lsn stamp at the start of the file */
	success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
	if (!success) {

		goto func_exit;
	}

	mach_write_to_8(page + FIL_PAGE_FILE_FLUSH_LSN, current_lsn);

	success = os_file_write(filepath, file, page, 0, 0, UNIV_PAGE_SIZE);
	if (!success) {

		goto func_exit;
	}
	success = os_file_flush(file);
func_exit:
	os_file_close(file);
	ut_free(page);
	ut_free(filepath);

	return(success);
}

unknown's avatar
unknown committed
2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435
/************************************************************************
Tries to open a single-table tablespace and checks the space id is right in
it. If does not succeed, prints an error message to the .err log. This
function is used to open the tablespace when we load a table definition
to the dictionary cache. NOTE that we assume this operation is used under the
protection of the dictionary mutex, so that two users cannot race here. This
operation does not leave the file associated with the tablespace open, but
closes it after we have looked at the space id in it. */

ibool
fil_open_single_table_tablespace(
/*=============================*/
			/* out: TRUE if success */
	ulint	id,	/* in: space id */
	char*	name)	/* in: table name in the databasename/tablename
			format */
{
	os_file_t	file;
	char*		filepath;
	ibool		success;
	byte*		page;
	ulint		space_id;
	ibool		ret		= TRUE;

	filepath = ut_malloc(OS_FILE_MAX_PATH);

	ut_a(strlen(name) < OS_FILE_MAX_PATH - 10);

unknown's avatar
unknown committed
2436
	sprintf(filepath, "%s/%s.ibd", fil_path_to_mysql_datadir, name);
unknown's avatar
unknown committed
2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452
					
	srv_normalize_path_for_win(filepath);

	file = os_file_create_simple_no_error_handling(filepath, OS_FILE_OPEN,
						OS_FILE_READ_ONLY, &success);
	if (!success) {
		/* The following call prints an error message */
		os_file_get_last_error(TRUE);

		ut_print_timestamp(stderr);

	        fprintf(stderr,
"  InnoDB: Error: trying to open a table, but could not\n"
"InnoDB: open the tablespace file %s!\n", filepath);
		fprintf(stderr,
"InnoDB: have you moved InnoDB .ibd files around without using the\n"
2453 2454 2455
"InnoDB: commands DISCARD TABLESPACE and IMPORT TABLESPACE?\n"
"InnoDB: You can look from section 15.1 of http://www.innodb.com/ibman.html\n"
"InnoDB: how to resolve the issue.\n");
unknown's avatar
unknown committed
2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479

		ut_free(filepath);

		return(FALSE);
	}

	/* Read the first page of the tablespace */

	page = ut_malloc(UNIV_PAGE_SIZE);

	success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);

	/* We have to read the tablespace id from the file */

	space_id = fsp_header_get_space_id(page);

	if (space_id != id) {
		ut_print_timestamp(stderr);

	        fprintf(stderr,
"  InnoDB: Error: tablespace id in file %s is %lu, but in the InnoDB\n"
"InnoDB: data dictionary it is %lu.\n", filepath, space_id, id);
		fprintf(stderr,
"InnoDB: Have you moved InnoDB .ibd files around without using the\n"
2480 2481 2482
"InnoDB: commands DISCARD TABLESPACE and IMPORT TABLESPACE?\n"
"InnoDB: You can look from section 15.1 of http://www.innodb.com/ibman.html\n"
"InnoDB: how to resolve the issue.\n");
unknown's avatar
unknown committed
2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525

		ret = FALSE;

		goto func_exit;
	}

	success = fil_space_create(filepath, space_id, FIL_TABLESPACE);

	if (!success) {
		goto func_exit;
	}

	/* We do not measure the size of the file, that is why we pass the 0
	below */

	fil_node_create(filepath, 0, space_id, FALSE);
func_exit:
	os_file_close(file);
	ut_free(page);
	ut_free(filepath);

	return(ret);
}

/************************************************************************
Opens an .ibd file and adds the associated single-table tablespace to the
InnoDB fil0fil.c data structures. */
static
void
fil_load_single_table_tablespace(
/*=============================*/
	char*	dbname,		/* in: database name */
	char*	filename)	/* in: file name (not a path), including the
				.ibd extension */
{
	os_file_t	file;
	char*		filepath;
	ibool		success;
	byte*		page;
	ulint		space_id;
	ulint		size_low;
	ulint		size_high;
	ib_longlong	size;
unknown's avatar
unknown committed
2526 2527 2528
#ifdef UNIV_HOTBACKUP
	fil_space_t*	space;
#endif
unknown's avatar
unknown committed
2529 2530
	filepath = ut_malloc(OS_FILE_MAX_PATH);

unknown's avatar
unknown committed
2531 2532
	ut_a(strlen(dbname) + strlen(filename) 
	+ strlen(fil_path_to_mysql_datadir) < OS_FILE_MAX_PATH - 100);
unknown's avatar
unknown committed
2533

unknown's avatar
unknown committed
2534 2535
	sprintf(filepath, "%s/%s/%s", fil_path_to_mysql_datadir, dbname,
								filename);
unknown's avatar
unknown committed
2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568
	srv_normalize_path_for_win(filepath);

	file = os_file_create_simple_no_error_handling(filepath, OS_FILE_OPEN,
						OS_FILE_READ_ONLY, &success);
	if (!success) {
		/* The following call prints an error message */
		os_file_get_last_error(TRUE);

	        fprintf(stderr,
"InnoDB: Error: could not open single-table tablespace file\n"
"InnoDB: %s!", filepath);

		ut_free(filepath);

		return;
	}

	success = os_file_get_size(file, &size_low, &size_high);

	if (!success) {
		/* The following call prints an error message */
		os_file_get_last_error(TRUE);

	        fprintf(stderr,
"InnoDB: Error: could not measure the size of single-table tablespace file\n"
"InnoDB: %s!", filepath);

		os_file_close(file);
		ut_free(filepath);

		return;
	}

unknown's avatar
unknown committed
2569 2570
	/* Every .ibd file is created >= 4 pages in size. Smaller files
	cannot be ok. */
unknown's avatar
unknown committed
2571

unknown's avatar
unknown committed
2572 2573 2574
	size = (((ib_longlong)size_high) << 32) + (ib_longlong)size_low;
#ifndef UNIV_HOTBACKUP
	if (size < FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
unknown's avatar
unknown committed
2575 2576 2577 2578 2579 2580 2581 2582 2583
	        fprintf(stderr,
"InnoDB: Error: the size of single-table tablespace file %s\n"
"InnoDB: is only %lu %lu, should be at least %lu!", filepath, size_high,
				       size_low, (ulint)4 * UNIV_PAGE_SIZE);
		os_file_close(file);
		ut_free(filepath);

		return;
	}
unknown's avatar
unknown committed
2584 2585
#endif
	/* Read the first page of the tablespace if the size big enough */
unknown's avatar
unknown committed
2586 2587 2588

	page = ut_malloc(UNIV_PAGE_SIZE);

unknown's avatar
unknown committed
2589 2590
	if (size >= FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
		success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
unknown's avatar
unknown committed
2591

unknown's avatar
unknown committed
2592
		/* We have to read the tablespace id from the file */
unknown's avatar
unknown committed
2593

unknown's avatar
unknown committed
2594 2595 2596 2597
		space_id = fsp_header_get_space_id(page);
	} else {
		space_id = ULINT_UNDEFINED;
	}
unknown's avatar
unknown committed
2598

unknown's avatar
unknown committed
2599
#ifndef UNIV_HOTBACKUP
unknown's avatar
unknown committed
2600 2601 2602 2603 2604 2605
	if (space_id == ULINT_UNDEFINED || space_id == 0) {
	        fprintf(stderr,
"InnoDB: Error: tablespace id %lu in file %s is not sensible\n", space_id,
								 filepath);
		goto func_exit;
	}
unknown's avatar
unknown committed
2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619
#else
	if (space_id == ULINT_UNDEFINED || space_id == 0) {
		char*	new_path;

		fprintf(stderr,
"InnoDB: Renaming tablespace %s of id %lu,\n"
"InnoDB: to %s_ibbackup_old_vers_<timestamp>\n"
"InnoDB: because its size %lld is too small (< 4 pages 16 kB each),\n"
"InnoDB: or the space id in the file header is not sensible.\n"
"InnoDB: This can happen in an ibbackup run, and is not dangerous.\n",
				filepath, space_id, filepath, size);
		os_file_close(file);

		new_path = ut_malloc(OS_FILE_MAX_PATH);
unknown's avatar
unknown committed
2620

unknown's avatar
unknown committed
2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672
		sprintf(new_path, "%s_ibbackup_old_vers_", filepath);
		ut_sprintf_timestamp_without_extra_chars(
					new_path + ut_strlen(new_path));
		ut_a(os_file_rename(filepath, new_path));

		ut_free(page);
		ut_free(filepath);
		ut_free(new_path);

		return;
	}

	/* A backup may contain the same space several times, if the space got
	renamed at a sensitive time. Since it is enough to have one version of
	the space, we rename the file if a space with the same space id
	already exists in the tablespace memory cache. We rather rename the
	file than delete it, because if there is a bug, we do not want to
	destroy valuable data. */

	mutex_enter(&(fil_system->mutex));

	space = fil_get_space_for_id_low(space_id);

	if (space) {
		char*	new_path;

		fprintf(stderr,
"InnoDB: Renaming tablespace %s of id %lu,\n"
"InnoDB: to %s_ibbackup_old_vers_<timestamp>\n"
"InnoDB: because space %s with the same id\n"
"InnoDB: was scanned earlier. This can happen if you have renamed tables\n"
"InnoDB: during an ibbackup run.\n", filepath, space_id, filepath,
								space->name);
		os_file_close(file);

		new_path = ut_malloc(OS_FILE_MAX_PATH);

		sprintf(new_path, "%s_ibbackup_old_vers_", filepath);
		ut_sprintf_timestamp_without_extra_chars(
					new_path + ut_strlen(new_path));
		mutex_exit(&(fil_system->mutex));

		ut_a(os_file_rename(filepath, new_path));

		ut_free(page);
		ut_free(filepath);
		ut_free(new_path);

		return;
	}
	mutex_exit(&(fil_system->mutex));
#endif
unknown's avatar
unknown committed
2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711
	success = fil_space_create(filepath, space_id, FIL_TABLESPACE);

	if (!success) {

		goto func_exit;
	}

	/* We do not measure the size of the file, that is why we pass the 0
	below */

	fil_node_create(filepath, 0, space_id, FALSE);
func_exit:
	os_file_close(file);
	ut_free(page);
	ut_free(filepath);
}

/************************************************************************
At the server startup, if we need crash recovery, scans the database
directories under the MySQL datadir, looking for .ibd files. Those files are
single-table tablespaces. We need to know the space id in each of them so that
we know into which file we should look to check the contents of a page stored
in the doublewrite buffer, also to know where to apply log records where the
space id is != 0. */

ulint
fil_load_single_table_tablespaces(void)
/*===================================*/
			/* out: DB_SUCCESS or error number */
{
	int		ret;
	char*		dbpath;
	os_file_dir_t	dir;
	os_file_dir_t	dbdir;
	os_file_stat_t	dbinfo;
	os_file_stat_t	fileinfo;

	/* The datadir of MySQL is always the default directory of mysqld */

unknown's avatar
unknown committed
2712
	dir = os_file_opendir(fil_path_to_mysql_datadir, TRUE);
unknown's avatar
unknown committed
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723

	if (dir == NULL) {

		return(DB_ERROR);
	}

	dbpath = ut_malloc(OS_FILE_MAX_PATH);

	/* Scan all directories under the datadir. They are the database
	directories of MySQL. */

unknown's avatar
unknown committed
2724 2725
	ret = os_file_readdir_next_file(fil_path_to_mysql_datadir, dir,
								&dbinfo);
unknown's avatar
unknown committed
2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
	while (ret == 0) {
		/* printf("Looking at %s in datadir\n", dbinfo.name); */

		if (dbinfo.type == OS_FILE_TYPE_FILE
		    || dbinfo.type == OS_FILE_TYPE_UNKNOWN) {

		        goto next_datadir_item;
		}

		/* We found a symlink or a directory; try opening it to see
		if a symlink is a directory */
		
		ut_a(strlen(dbinfo.name) < OS_FILE_MAX_PATH - 10);

unknown's avatar
unknown committed
2740 2741
		sprintf(dbpath, "%s/%s", fil_path_to_mysql_datadir,
								dbinfo.name);
unknown's avatar
unknown committed
2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784
		srv_normalize_path_for_win(dbpath);

		dbdir = os_file_opendir(dbpath, FALSE);

		if (dbdir != NULL) {
			/* printf("Opened dir %s\n", dbinfo.name); */

			/* We found a database directory; loop through it,
			looking for possible .ibd files in it */

			ret = os_file_readdir_next_file(dbpath, dbdir,
								&fileinfo);
			while (ret == 0) {
				/* printf(
"     Looking at file %s\n", fileinfo.name); */

			        if (fileinfo.type == OS_FILE_TYPE_DIR
				    || dbinfo.type == OS_FILE_TYPE_UNKNOWN) {
				        goto next_file_item;
				}

				/* We found a symlink or a file */
				if (strlen(fileinfo.name) > 4
				    && 0 == strcmp(fileinfo.name + 
						strlen(fileinfo.name) - 4,
						".ibd")) {
				        /* The name ends in .ibd; try opening
					the file */
					fil_load_single_table_tablespace(
						dbinfo.name, fileinfo.name);
				}
next_file_item:
				ret = os_file_readdir_next_file(dbpath, dbdir,
								&fileinfo);
			}

			if (0 != os_file_closedir(dbdir)) {
				 fprintf(stderr,
"InnoDB: Warning: could not close database directory %s\n", dbpath);
			}
		}
		
next_datadir_item:
unknown's avatar
unknown committed
2785 2786
		ret = os_file_readdir_next_file(fil_path_to_mysql_datadir,
								dir, &dbinfo);
unknown's avatar
unknown committed
2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834
	}

	ut_free(dbpath);

	/* At the end of directory we should get 1 as the return value, -1
	if there was an error */
	if (ret != 1) {
		fprintf(stderr,
"InnoDB: Error: os_file_readdir_next_file returned %d in MySQL datadir\n",
							       ret);
		os_file_closedir(dir);

		return(DB_ERROR);
	}

	if (0 != os_file_closedir(dir)) {
		fprintf(stderr,
"InnoDB: Error: could not close MySQL datadir\n");

		return(DB_ERROR);
	}

	return(DB_SUCCESS);
}

/************************************************************************
If we need crash recovery, and we have called
fil_load_single_table_tablespaces() and dict_load_single_table_tablespaces(),
we can call this function to print an error message of orphaned .ibd files
for which there is not a data dictionary entry with a matching table name
and space id. */

void
fil_print_orphaned_tablespaces(void)
/*================================*/
{
	fil_system_t*	system 		= fil_system;
	fil_space_t*	space;

	mutex_enter(&(system->mutex));

	space = UT_LIST_GET_FIRST(system->space_list);

	while (space) {
	        if (space->purpose == FIL_TABLESPACE && space->id != 0
							  && !space->mark) {
			fprintf(stderr,
"InnoDB: Warning: tablespace %s of id %lu has no matching table in\n"
unknown's avatar
unknown committed
2835
"InnoDB: the InnoDB data dictionary.\n", space->name, space->id);
unknown's avatar
unknown committed
2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
		}

		space = UT_LIST_GET_NEXT(space_list, space);
	}

	mutex_exit(&(system->mutex));	
}

/***********************************************************************
Returns TRUE if a single-table tablespace does not exist in the memory cache,
or is being deleted there. */

ibool
fil_tablespace_deleted_or_being_deleted_in_mem(
/*===========================================*/
				/* out: TRUE if does not exist or is being\
				deleted */
	ulint		id,	/* in: space id */
	ib_longlong	version)/* in: tablespace_version should be this; if
				you pass -1 as the value of this, then this
				parameter is ignored */
{
	fil_system_t*	system	= fil_system;
	fil_space_t*	space;

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL || space->is_being_deleted) {
		mutex_exit(&(system->mutex));

		return(TRUE);
	}

unknown's avatar
unknown committed
2873 2874
	if (version != ((ib_longlong)-1)
				&& space->tablespace_version != version) {
unknown's avatar
unknown committed
2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921
		mutex_exit(&(system->mutex));

		return(TRUE);
	}

	mutex_exit(&(system->mutex));

	return(FALSE);
}

/***********************************************************************
Returns TRUE if a single-table tablespace exists in the memory cache. */

ibool
fil_tablespace_exists_in_mem(
/*=========================*/
			/* out: TRUE if exists */
	ulint	id)	/* in: space id */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	space;

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	if (space == NULL) {
		mutex_exit(&(system->mutex));

		return(FALSE);
	}

	mutex_exit(&(system->mutex));

	return(TRUE);
}

/***********************************************************************
Returns TRUE if a matching tablespace exists in the InnoDB tablespace memory
cache. Note that if we have not done a crash recovery at the database startup,
there may be many tablespaces which are not yet in the memory cache. */

ibool
fil_space_for_table_exists_in_mem(
/*==============================*/
unknown's avatar
unknown committed
2922 2923
				/* out: TRUE if a matching tablespace exists
				in the memory cache */
unknown's avatar
unknown committed
2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
	ulint	id,		/* in: space id */
	char*	name,		/* in: table name in the standard
				'databasename/tablename' format */
	ibool	mark_space,	/* in: in crash recovery, at database startup
				we mark all spaces which have an associated
				table in the InnoDB data dictionary, so that
				we can print a warning about orphaned
				tablespaces */
	ibool	print_error_if_does_not_exist)
				/* in: print detailed error information to
				the .err log if a matching tablespace is
				not found from memory */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	namespace;
	fil_space_t*	space;
	char		path[OS_FILE_MAX_PATH];

	ut_ad(system);

	mutex_enter(&(system->mutex));

unknown's avatar
unknown committed
2946
	sprintf(path, "%s/%s.ibd", fil_path_to_mysql_datadir, name);
unknown's avatar
unknown committed
2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958
	srv_normalize_path_for_win(path);

	/* Look if there is a space with the same id */

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

	/* Look if there is a space with the same name; the name is the
	directory path from the datadir to the file */

	HASH_SEARCH(name_hash, system->name_hash,
					ut_fold_string(path), namespace,
					0 == strcmp(namespace->name, path));
unknown's avatar
unknown committed
2959 2960 2961 2962 2963 2964
	if (space && space == namespace) {
		/* Found */
		
		if (mark_space) {
			space->mark = TRUE;
		}
unknown's avatar
unknown committed
2965

unknown's avatar
unknown committed
2966
		mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
2967

unknown's avatar
unknown committed
2968 2969 2970 2971
		return(TRUE);
	}

	if (!print_error_if_does_not_exist) {
unknown's avatar
unknown committed
2972 2973 2974 2975 2976 2977 2978 2979 2980 2981
		
		mutex_exit(&(system->mutex));
		
		return(FALSE);
	}

	if (space == NULL) {
		if (namespace == NULL) {
		        ut_print_timestamp(stderr);
			fprintf(stderr,
unknown's avatar
unknown committed
2982 2983 2984 2985 2986
"  InnoDB: Error: table %s\n"
"InnoDB: in InnoDB data dictionary has tablespace id %lu,\n"
"InnoDB: but tablespace with that id or name does not exist. Have\n"
"InnoDB: you deleted or moved .ibd files?\n",
			       name, id);
unknown's avatar
unknown committed
2987 2988 2989
		} else {
		        ut_print_timestamp(stderr);
			fprintf(stderr,
unknown's avatar
unknown committed
2990 2991 2992
"  InnoDB: Error: table %s\n"
"InnoDB: in InnoDB data dictionary has tablespace id %lu,\n"
"InnoDB: but tablespace with that id does not exist. There is\n"
unknown's avatar
unknown committed
2993
"InnoDB: a tablespace of name %s and id %lu, though. Have\n"
unknown's avatar
unknown committed
2994 2995
"InnoDB: you deleted or moved .ibd files?\n",
			       name, id, namespace->name, namespace->id);
unknown's avatar
unknown committed
2996
		}
unknown's avatar
unknown committed
2997 2998 2999
		fprintf(stderr,
"InnoDB: You can look from section 15.1 of http://www.innodb.com/ibman.html\n"
"InnoDB: how to resolve the issue.\n");
unknown's avatar
unknown committed
3000 3001 3002 3003 3004 3005 3006 3007 3008

		mutex_exit(&(system->mutex));

		return(FALSE);
	}

	if (0 != strcmp(space->name, path)) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
unknown's avatar
unknown committed
3009 3010 3011
"  InnoDB: Error: table %s\n"
"InnoDB: in InnoDB data dictionary has tablespace id %lu,\n"
"InnoDB: but tablespace with that id has name %s.\n"
unknown's avatar
unknown committed
3012 3013
"InnoDB: Have you deleted or moved .ibd files?\n", name, id, space->name);

unknown's avatar
unknown committed
3014 3015
		if (namespace != NULL) {
			fprintf(stderr,
unknown's avatar
unknown committed
3016 3017
"InnoDB: There is a tablespace with the right name\n"
"InnoDB: %s, but its id is %lu.\n", namespace->name, namespace->id);
unknown's avatar
unknown committed
3018 3019
		}

unknown's avatar
unknown committed
3020 3021 3022 3023
		fprintf(stderr,
"InnoDB: You can look from section 15.1 of http://www.innodb.com/ibman.html\n"
"InnoDB: how to resolve the issue.\n");

unknown's avatar
unknown committed
3024 3025 3026 3027 3028
		mutex_exit(&(system->mutex));

		return(FALSE);
	}

unknown's avatar
unknown committed
3029
	mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
3030

unknown's avatar
unknown committed
3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066
	return(FALSE);
}

/***********************************************************************
Checks if a single-table tablespace for a given table name exists in the
tablespace memory cache. */
static
ulint
fil_get_space_id_for_table(
/*=======================*/
				/* out: space id, ULINT_UNDEFINED if not
				found */
	char*	name)		/* in: table name in the standard
				'databasename/tablename' format */
{
	fil_system_t*	system		= fil_system;
	fil_space_t*	namespace;
	ulint		id		= ULINT_UNDEFINED;
	char		path[OS_FILE_MAX_PATH];

	ut_ad(system);

	mutex_enter(&(system->mutex));

	sprintf(path, "%s/%s.ibd", fil_path_to_mysql_datadir, name);
	srv_normalize_path_for_win(path);

	/* Look if there is a space with the same name; the name is the
	directory path to the file */

	HASH_SEARCH(name_hash, system->name_hash,
					ut_fold_string(path), namespace,
					0 == strcmp(namespace->name, path));
	if (namespace) {
		id = namespace->id;
	}	
unknown's avatar
unknown committed
3067 3068 3069

	mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
3070
	return(id);
unknown's avatar
unknown committed
3071 3072 3073
}

/**************************************************************************
unknown's avatar
unknown committed
3074 3075 3076
Tries to extend a data file so that it would accommodate the number of pages
given. The tablespace must be cached in the memory cache. If the space is big
enough already, does nothing. */
unknown's avatar
unknown committed
3077 3078

ibool
unknown's avatar
unknown committed
3079 3080 3081 3082 3083 3084 3085 3086 3087 3088
fil_extend_space_to_desired_size(
/*=============================*/
				/* out: TRUE if success */
	ulint*	actual_size,	/* out: size of the space after extension;
				if we ran out of disk space this may be lower
				than the desired size */
	ulint	space_id,	/* in: space id, must be != 0 */
	ulint	size_after_extend)/* in: desired size in pages after the
				extension; if the current space size is bigger
				than this already, the function does nothing */
unknown's avatar
unknown committed
3089 3090 3091 3092 3093 3094
{
	fil_system_t*	system		= fil_system;
	fil_node_t*	node;
	fil_space_t*	space;
	byte*		buf2;
	byte*		buf;
unknown's avatar
unknown committed
3095 3096 3097 3098 3099 3100
	ulint		start_page_no;
	ulint		file_start_page_no;
	ulint		n_pages;
	ulint		offset_high;
	ulint		offset_low;
	ibool		success		= TRUE;
unknown's avatar
unknown committed
3101 3102 3103 3104 3105 3106

	fil_mutex_enter_and_prepare_for_io(space_id);

	HASH_SEARCH(hash, system->spaces, space_id, space,
						space->id == space_id);
	ut_a(space);
unknown's avatar
unknown committed
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116

	if (space->size >= size_after_extend) {
		/* Space already big enough */

		*actual_size = space->size;

		mutex_exit(&(system->mutex));	

		return(TRUE);
	}
unknown's avatar
unknown committed
3117 3118 3119 3120 3121
	
	node = UT_LIST_GET_LAST(space->chain);

	fil_node_prepare_for_io(node, system, space);

unknown's avatar
unknown committed
3122
	/* Extend 1 MB at a time */
unknown's avatar
unknown committed
3123 3124 3125 3126 3127 3128

	buf2 = mem_alloc(1024 * 1024 + UNIV_PAGE_SIZE);
	buf = ut_align(buf2, UNIV_PAGE_SIZE);

	memset(buf, '\0', 1024 * 1024);

unknown's avatar
unknown committed
3129 3130 3131 3132 3133 3134 3135 3136 3137
	start_page_no = space->size;
	file_start_page_no = space->size - node->size;

	while (start_page_no < size_after_extend) {	
		n_pages = size_after_extend - start_page_no;

		if (n_pages > (1024 * 1024) / UNIV_PAGE_SIZE) {
			n_pages = (1024 * 1024) / UNIV_PAGE_SIZE;
		}
unknown's avatar
unknown committed
3138

unknown's avatar
unknown committed
3139 3140 3141 3142 3143 3144 3145 3146 3147 3148
		offset_high = (start_page_no - file_start_page_no)
				/ (4096 * ((1024 * 1024) / UNIV_PAGE_SIZE));
		offset_low  = ((start_page_no - file_start_page_no)
				% (4096 * ((1024 * 1024) / UNIV_PAGE_SIZE)))
			      * UNIV_PAGE_SIZE;
#ifdef UNIV_HOTBACKUP
		success = os_file_write(node->name, node->handle, buf,
					offset_low, offset_high,
					UNIV_PAGE_SIZE * n_pages);
#else
unknown's avatar
unknown committed
3149 3150
		success = os_aio(OS_FILE_WRITE, OS_AIO_SYNC,
			node->name, node->handle, buf,
unknown's avatar
unknown committed
3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169
			offset_low, offset_high,
			UNIV_PAGE_SIZE * n_pages,
			NULL, NULL);
#endif
		if (success) {
			node->size += n_pages;
			space->size += n_pages;

			os_has_said_disk_full = FALSE;
		} else {
			/* Let us measure the size of the file to determine
			how much we were able to extend it */
			
			n_pages = ((ulint)
				(os_file_get_size_as_iblonglong(node->handle)
				/ UNIV_PAGE_SIZE)) - node->size;

			node->size += n_pages;
			space->size += n_pages;
unknown's avatar
unknown committed
3170 3171 3172 3173

			break;
		}

unknown's avatar
unknown committed
3174
		start_page_no += n_pages;
unknown's avatar
unknown committed
3175 3176 3177 3178 3179 3180
	}

	mem_free(buf2);

	fil_node_complete_io(node, system, OS_FILE_WRITE);

unknown's avatar
unknown committed
3181 3182 3183 3184
	*actual_size = space->size;
	/*
        printf("Extended %s to %lu, actual size %lu pages\n", space->name,
                                        size_after_extend, *actual_size); */
unknown's avatar
unknown committed
3185 3186 3187 3188
	mutex_exit(&(system->mutex));	

	fil_flush(space_id);

unknown's avatar
unknown committed
3189
	return(success);
3190 3191
}

unknown's avatar
unknown committed
3192 3193 3194 3195 3196 3197
#ifdef UNIV_HOTBACKUP
/************************************************************************
Extends all tablespaces to the size stored in the space header. During the
ibbackup --apply-log phase we extended the spaces on-demand so that log records
could be appllied, but that may have left spaces still too small compared to
the size stored in the space header. */
3198

unknown's avatar
unknown committed
3199 3200 3201
void
fil_extend_tablespaces_to_stored_len(void)
/*======================================*/
3202
{
unknown's avatar
unknown committed
3203
	fil_system_t*	system 		= fil_system;
unknown's avatar
unknown committed
3204 3205
	fil_space_t*	space;
	byte*		buf;
unknown's avatar
unknown committed
3206 3207 3208
	ulint		actual_size;
	ulint		size_in_header;
	ulint		error;
unknown's avatar
unknown committed
3209 3210
	ibool		success;

unknown's avatar
unknown committed
3211
	buf = mem_alloc(UNIV_PAGE_SIZE);
unknown's avatar
unknown committed
3212

unknown's avatar
unknown committed
3213
	mutex_enter(&(system->mutex));
unknown's avatar
unknown committed
3214

unknown's avatar
unknown committed
3215
	space = UT_LIST_GET_FIRST(system->space_list);
3216

unknown's avatar
unknown committed
3217 3218
	while (space) {
	        ut_a(space->purpose == FIL_TABLESPACE);
3219

unknown's avatar
unknown committed
3220 3221 3222 3223 3224 3225
		mutex_exit(&(system->mutex)); /* no need to protect with a
					      mutex, because this is a single-
					      threaded operation */
		error = fil_read(TRUE, space->id, 0, 0, UNIV_PAGE_SIZE, buf,
									NULL);
		ut_a(error == DB_SUCCESS);
3226

unknown's avatar
unknown committed
3227
		size_in_header = fsp_get_size_low(buf);
3228

unknown's avatar
unknown committed
3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239
		success = fil_extend_space_to_desired_size(&actual_size,
						space->id, size_in_header);
		if (!success) {
			fprintf(stderr,
"InnoDB: Error: could not extend the tablespace of %s\n"
"InnoDB: to the size stored in header, %lu pages;\n"
"InnoDB: size after extension %lu pages\n"
"InnoDB: Check that you have free disk space and retry!\n", space->name,
					size_in_header, actual_size);
			exit(1);				
		}
unknown's avatar
unknown committed
3240

unknown's avatar
unknown committed
3241
		mutex_enter(&(system->mutex));
unknown's avatar
unknown committed
3242

unknown's avatar
unknown committed
3243
		space = UT_LIST_GET_NEXT(space_list, space);
3244 3245
	}

unknown's avatar
unknown committed
3246
	mutex_exit(&(system->mutex));
unknown's avatar
unknown committed
3247

unknown's avatar
unknown committed
3248
	mem_free(buf);
3249
}
unknown's avatar
unknown committed
3250
#endif
3251

unknown's avatar
unknown committed
3252 3253
/*========== RESERVE FREE EXTENTS (for a B-tree split, for example) ===*/

3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265
/***********************************************************************
Tries to reserve free extents in a file space. */

ibool
fil_space_reserve_free_extents(
/*===========================*/
				/* out: TRUE if succeed */
	ulint	id,		/* in: space id */
	ulint	n_free_now,	/* in: number of free extents now */
	ulint	n_to_reserve)	/* in: how many one wants to reserve */
{
	fil_system_t*	system		= fil_system;
unknown's avatar
unknown committed
3266
	fil_space_t*	space;
3267 3268 3269 3270 3271 3272 3273 3274
	ibool		success;

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
3275 3276
	ut_a(space);

3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298
	if (space->n_reserved_extents + n_to_reserve > n_free_now) {
		success = FALSE;
	} else {
		space->n_reserved_extents += n_to_reserve;
		success = TRUE;
	}
	
	mutex_exit(&(system->mutex));

	return(success);
}

/***********************************************************************
Releases free extents in a file space. */

void
fil_space_release_free_extents(
/*===========================*/
	ulint	id,		/* in: space id */
	ulint	n_reserved)	/* in: how many one reserved */
{
	fil_system_t*	system		= fil_system;
unknown's avatar
unknown committed
3299
	fil_space_t*	space;
3300 3301 3302 3303 3304 3305 3306

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);

unknown's avatar
unknown committed
3307
	ut_a(space);
3308 3309 3310 3311 3312 3313 3314
	ut_a(space->n_reserved_extents >= n_reserved);
	
	space->n_reserved_extents -= n_reserved;
	
	mutex_exit(&(system->mutex));
}

3315 3316 3317 3318 3319 3320 3321 3322 3323 3324
/***********************************************************************
Gets the number of reserved extents. If the database is silent, this number
should be zero. */

ulint
fil_space_get_n_reserved_extents(
/*=============================*/
	ulint	id)		/* in: space id */
{
	fil_system_t*	system		= fil_system;
unknown's avatar
unknown committed
3325
	fil_space_t*	space;
3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
	ulint		n;

	ut_ad(system);

	mutex_enter(&(system->mutex));

	HASH_SEARCH(hash, system->spaces, id, space, space->id == id);
	
	ut_a(space);

	n = space->n_reserved_extents;
	
	mutex_exit(&(system->mutex));

	return(n);
}

unknown's avatar
unknown committed
3343 3344
/*============================ FILE I/O ================================*/

3345
/************************************************************************
unknown's avatar
unknown committed
3346 3347
NOTE: you must call fil_mutex_enter_and_prepare_for_io() first!

3348 3349
Prepares a file node for i/o. Opens the file if it is closed. Updates the
pending i/o's field in the node and the system appropriately. Takes the node
unknown's avatar
unknown committed
3350 3351
off the LRU list if it is in the LRU list. The caller must hold the fil_sys
mutex. */
3352 3353 3354 3355 3356
static
void
fil_node_prepare_for_io(
/*====================*/
	fil_node_t*	node,	/* in: file node */
unknown's avatar
unknown committed
3357
	fil_system_t*	system,	/* in: tablespace memory cache */
3358 3359 3360 3361 3362
	fil_space_t*	space)	/* in: space */
{
	ut_ad(node && system && space);
	ut_ad(mutex_own(&(system->mutex)));
	
unknown's avatar
unknown committed
3363 3364 3365 3366 3367 3368 3369
	if (system->n_open > system->max_n_open + 5) {
		ut_print_timestamp(stderr);
		fprintf(stderr,
"  InnoDB: Warning: open files %lu exceeds the limit %lu\n",
					system->n_open, system->max_n_open);
	}

3370
	if (node->open == FALSE) {
unknown's avatar
unknown committed
3371
		/* File is closed: open it */
3372 3373
		ut_a(node->n_pending == 0);

unknown's avatar
unknown committed
3374
		fil_node_open_file(node, system, space);
3375 3376
	}

unknown's avatar
unknown committed
3377 3378
	if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
						      && space->id != 0) {
3379 3380
		/* The node is in the LRU list, remove it */

unknown's avatar
unknown committed
3381
		ut_a(UT_LIST_GET_LEN(system->LRU) > 0);
3382

unknown's avatar
unknown committed
3383
		UT_LIST_REMOVE(LRU, system->LRU, node);
3384
	}
unknown's avatar
unknown committed
3385 3386

	node->n_pending++;
3387 3388 3389 3390
}

/************************************************************************
Updates the data structures when an i/o operation finishes. Updates the
unknown's avatar
unknown committed
3391
pending i/o's field in the node appropriately. */
3392 3393 3394 3395 3396
static
void
fil_node_complete_io(
/*=================*/
	fil_node_t*	node,	/* in: file node */
unknown's avatar
unknown committed
3397 3398 3399 3400
	fil_system_t*	system,	/* in: tablespace memory cache */
	ulint		type)	/* in: OS_FILE_WRITE or OS_FILE_READ; marks
				the node as modified if
				type == OS_FILE_WRITE */
3401 3402 3403 3404
{
	ut_ad(node);
	ut_ad(system);
	ut_ad(mutex_own(&(system->mutex)));
unknown's avatar
unknown committed
3405

3406 3407 3408 3409
	ut_a(node->n_pending > 0);
	
	node->n_pending--;

unknown's avatar
unknown committed
3410 3411 3412
	if (type == OS_FILE_WRITE) {
		system->modification_counter++;
		node->modification_counter = system->modification_counter;
3413 3414
	}
	
unknown's avatar
unknown committed
3415 3416
	if (node->n_pending == 0 && node->space->purpose == FIL_TABLESPACE
					&& node->space->id != 0) {
3417 3418
		/* The node must be put back to the LRU list */
		UT_LIST_ADD_FIRST(LRU, system->LRU, node);
unknown's avatar
unknown committed
3419 3420 3421
	}
}

3422 3423 3424
/************************************************************************
Reads or writes data. This operation is asynchronous (aio). */

unknown's avatar
unknown committed
3425
ulint
3426 3427
fil_io(
/*===*/
unknown's avatar
unknown committed
3428 3429 3430
				/* out: DB_SUCCESS, or DB_TABLESPACE_DELETED
				if we are trying to do i/o on a tablespace
				which does not exist */
3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445
	ulint	type,		/* in: OS_FILE_READ or OS_FILE_WRITE,
				ORed to OS_FILE_LOG, if a log i/o
				and ORed to OS_AIO_SIMULATED_WAKE_LATER
				if simulated aio and we want to post a
				batch of i/os; NOTE that a simulated batch
				may introduce hidden chances of deadlocks,
				because i/os are not actually handled until
				all have been posted: use with great
				caution! */
	ibool	sync,		/* in: TRUE if synchronous aio is desired */
	ulint	space_id,	/* in: space id */
	ulint	block_offset,	/* in: offset in number of blocks */
	ulint	byte_offset,	/* in: remainder of offset in bytes; in
				aio this must be divisible by the OS block
				size */
unknown's avatar
unknown committed
3446 3447 3448
	ulint	len,		/* in: how many bytes to read or write; this
				must not cross a file boundary; in aio this
				must be a block size multiple */
3449 3450 3451 3452 3453 3454
	void*	buf,		/* in/out: buffer where to store read data
				or from where to write; in aio this must be
				appropriately aligned */
	void*	message)	/* in: message for aio handler if non-sync
				aio used, else ignored */
{
unknown's avatar
unknown committed
3455
	fil_system_t*	system		= fil_system;
3456 3457 3458 3459 3460 3461 3462 3463
	ulint		mode;
	fil_space_t*	space;
	fil_node_t*	node;
	ulint		offset_high;
	ulint		offset_low;
	ibool		ret;
	ulint		is_log;
	ulint		wake_later;
3464
	
3465 3466 3467 3468 3469 3470 3471 3472 3473
	is_log = type & OS_FILE_LOG;
	type = type & ~OS_FILE_LOG;

	wake_later = type & OS_AIO_SIMULATED_WAKE_LATER;
	type = type & ~OS_AIO_SIMULATED_WAKE_LATER;
	
	ut_ad(byte_offset < UNIV_PAGE_SIZE);
	ut_ad(buf);
	ut_ad(len > 0);
unknown's avatar
unknown committed
3474
	ut_a((1 << UNIV_PAGE_SIZE_SHIFT) == UNIV_PAGE_SIZE);
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486
	ut_ad(fil_validate());
#ifndef UNIV_LOG_DEBUG
	/* ibuf bitmap pages must be read in the sync aio mode: */
	ut_ad(recv_no_ibuf_operations || (type == OS_FILE_WRITE)
		|| !ibuf_bitmap_page(block_offset) || sync || is_log);
#ifdef UNIV_SYNC_DEBUG
	ut_ad(!ibuf_inside() || is_log || (type == OS_FILE_WRITE)
					|| ibuf_page(space_id, block_offset));
#endif
#endif
	if (sync) {
		mode = OS_AIO_SYNC;
3487
	} else if (type == OS_FILE_READ && !is_log
3488 3489 3490 3491 3492 3493 3494 3495
				&& ibuf_page(space_id, block_offset)) {
		mode = OS_AIO_IBUF;
	} else if (is_log) {
		mode = OS_AIO_LOG;
	} else {
		mode = OS_AIO_NORMAL;
	}

unknown's avatar
unknown committed
3496 3497
	/* Reserve the fil_system mutex and make sure that we can open at
	least one file while holding it, if the file is not already open */
3498

unknown's avatar
unknown committed
3499
	fil_mutex_enter_and_prepare_for_io(space_id);
3500
	
unknown's avatar
unknown committed
3501 3502 3503
	HASH_SEARCH(hash, system->spaces, space_id, space,
							space->id == space_id);
	if (!space) {
3504 3505
		mutex_exit(&(system->mutex));

unknown's avatar
unknown committed
3506
		ut_print_timestamp(stderr);
3507
		fprintf(stderr,
unknown's avatar
unknown committed
3508 3509 3510
"  InnoDB: Error: trying to do i/o to a tablespace which does not exist.\n"
"InnoDB: i/o type %lu, space id %lu, page no. %lu, i/o length %lu bytes\n",
		       type, space_id, block_offset, len);
3511

unknown's avatar
unknown committed
3512 3513
		return(DB_TABLESPACE_DELETED);
	}
3514 3515 3516 3517 3518 3519

	ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));

	node = UT_LIST_GET_FIRST(space->chain);

	for (;;) {
unknown's avatar
unknown committed
3520 3521 3522 3523 3524 3525 3526
		if (space->id != 0 && node->size == 0) {
			/* We do not know the size of a single-table tablespace
			before we open the file */

			break;
		}

unknown's avatar
unknown committed
3527 3528
		if (node == NULL) {
			fprintf(stderr,
unknown's avatar
unknown committed
3529 3530
	"InnoDB: Error: trying to access page number %lu in space %lu,\n"
	"InnoDB: space name %s,\n"
unknown's avatar
unknown committed
3531 3532
	"InnoDB: which is outside the tablespace bounds.\n"
	"InnoDB: Byte offset %lu, len %lu, i/o type %lu\n", 
unknown's avatar
unknown committed
3533 3534
 			block_offset, space_id, space->name, byte_offset, len,
			type);
unknown's avatar
unknown committed
3535 3536 3537
 			
			ut_a(0);
		}
3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550

		if (node->size > block_offset) {
			/* Found! */
			break;
		} else {
			block_offset -= node->size;
			node = UT_LIST_GET_NEXT(chain, node);
		}
	}		
	
	/* Open file if closed */
	fil_node_prepare_for_io(node, system, space);

unknown's avatar
unknown committed
3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565
	/* Check that at least the start offset is within the bounds of a
	single-table tablespace */
	if (space->purpose == FIL_TABLESPACE && space->id != 0
	    && node->size <= block_offset) {

	        fprintf(stderr,
	"InnoDB: Error: trying to access page number %lu in space %lu,\n"
	"InnoDB: space name %s,\n"
	"InnoDB: which is outside the tablespace bounds.\n"
	"InnoDB: Byte offset %lu, len %lu, i/o type %lu\n", 
 			block_offset, space_id, space->name, byte_offset, len,
			type);
 		ut_a(0);
	}

3566 3567 3568 3569 3570 3571
	/* Now we have made the changes in the data structures of system */
	mutex_exit(&(system->mutex));

	/* Calculate the low 32 bits and the high 32 bits of the file offset */

	offset_high = (block_offset >> (32 - UNIV_PAGE_SIZE_SHIFT));
unknown's avatar
unknown committed
3572
	offset_low  = ((block_offset << UNIV_PAGE_SIZE_SHIFT) & 0xFFFFFFFFUL)
3573 3574 3575 3576 3577 3578 3579
			+ byte_offset;

	ut_a(node->size - block_offset >=
 		(byte_offset + len + (UNIV_PAGE_SIZE - 1)) / UNIV_PAGE_SIZE);

	/* Do aio */

unknown's avatar
unknown committed
3580 3581
	ut_a(byte_offset % OS_FILE_LOG_BLOCK_SIZE == 0);
	ut_a((len % OS_FILE_LOG_BLOCK_SIZE) == 0);
3582

unknown's avatar
unknown committed
3583 3584 3585 3586 3587 3588 3589 3590 3591 3592
#ifdef UNIV_HOTBACKUP
	/* In ibbackup do normal i/o, not aio */
	if (type == OS_FILE_READ) {
		ret = os_file_read(node->handle, buf, offset_low, offset_high,
									len);
	} else {
		ret = os_file_write(node->name, node->handle, buf,
					offset_low, offset_high, len);
	}
#else
3593 3594 3595
	/* Queue the aio request */
	ret = os_aio(type, mode | wake_later, node->name, node->handle, buf,
				offset_low, offset_high, len, node, message);
unknown's avatar
unknown committed
3596
#endif
3597 3598 3599 3600 3601 3602 3603 3604
	ut_a(ret);

	if (mode == OS_AIO_SYNC) {
		/* The i/o operation is already completed when we return from
		os_aio: */
		
		mutex_enter(&(system->mutex));

3605
		fil_node_complete_io(node, system, type);
3606 3607 3608 3609 3610

		mutex_exit(&(system->mutex));

		ut_ad(fil_validate());
	}
unknown's avatar
unknown committed
3611 3612

	return(DB_SUCCESS);
3613 3614 3615 3616 3617 3618 3619
}

/************************************************************************
Reads data from a space to a buffer. Remember that the possible incomplete
blocks at the end of file are ignored: they are not taken into account when
calculating the byte offset within a space. */

unknown's avatar
unknown committed
3620
ulint
3621 3622
fil_read(
/*=====*/
unknown's avatar
unknown committed
3623 3624 3625
				/* out: DB_SUCCESS, or DB_TABLESPACE_DELETED
				if we are trying to do i/o on a tablespace
				which does not exist */
3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638
	ibool	sync,		/* in: TRUE if synchronous aio is desired */
	ulint	space_id,	/* in: space id */
	ulint	block_offset,	/* in: offset in number of blocks */
	ulint	byte_offset,	/* in: remainder of offset in bytes; in aio
				this must be divisible by the OS block size */
	ulint	len,		/* in: how many bytes to read; this must not
				cross a file boundary; in aio this must be a
				block size multiple */
	void*	buf,		/* in/out: buffer where to store data read;
				in aio this must be appropriately aligned */
	void*	message)	/* in: message for aio handler if non-sync
				aio used, else ignored */
{
unknown's avatar
unknown committed
3639 3640
	return(fil_io(OS_FILE_READ, sync, space_id, block_offset,
					  byte_offset, len, buf, message));
3641 3642 3643 3644 3645 3646 3647
}

/************************************************************************
Writes data to a space from a buffer. Remember that the possible incomplete
blocks at the end of file are ignored: they are not taken into account when
calculating the byte offset within a space. */

unknown's avatar
unknown committed
3648
ulint
3649 3650
fil_write(
/*======*/
unknown's avatar
unknown committed
3651 3652 3653
				/* out: DB_SUCCESS, or DB_TABLESPACE_DELETED
				if we are trying to do i/o on a tablespace
				which does not exist */
3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666
	ibool	sync,		/* in: TRUE if synchronous aio is desired */
	ulint	space_id,	/* in: space id */
	ulint	block_offset,	/* in: offset in number of blocks */
	ulint	byte_offset,	/* in: remainder of offset in bytes; in aio
				this must be divisible by the OS block size */
	ulint	len,		/* in: how many bytes to write; this must
				not cross a file boundary; in aio this must
				be a block size multiple */
	void*	buf,		/* in: buffer from which to write; in aio
				this must be appropriately aligned */
	void*	message)	/* in: message for aio handler if non-sync
				aio used, else ignored */
{
unknown's avatar
unknown committed
3667 3668
	return(fil_io(OS_FILE_WRITE, sync, space_id, block_offset,
					   byte_offset, len, buf, message));
3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682
}

/**************************************************************************
Waits for an aio operation to complete. This function is used to write the
handler for completed requests. The aio array of pending requests is divided
into segments (see os0file.c for more info). The thread specifies which
segment it wants to wait for. */

void
fil_aio_wait(
/*=========*/
	ulint	segment)	/* in: the number of the segment in the aio
				array to wait for */ 
{
unknown's avatar
unknown committed
3683
	fil_system_t*	system		= fil_system;
3684 3685 3686
	ibool		ret;		
	fil_node_t*	fil_node;
	void*		message;
3687
	ulint		type;
3688 3689 3690 3691
	
	ut_ad(fil_validate());

	if (os_aio_use_native_aio) {
unknown's avatar
unknown committed
3692
		srv_io_thread_op_info[segment] = (char *) "handle native aio";
3693
#ifdef WIN_ASYNC_IO
3694 3695
		ret = os_aio_windows_handle(segment, 0, &fil_node, &message,
								&type);
3696 3697 3698
#elif defined(POSIX_ASYNC_IO)
		ret = os_aio_posix_handle(segment, &fil_node, &message);
#else
3699
		ret = 0; /* Eliminate compiler warning */
3700 3701 3702
		ut_a(0);
#endif
	} else {
unknown's avatar
unknown committed
3703
		srv_io_thread_op_info[segment] =(char *)"handle simulated aio";
3704

3705
		ret = os_aio_simulated_handle(segment, (void**) &fil_node,
3706
	                                               &message, &type);
3707 3708 3709
	}
	
	ut_a(ret);
3710

unknown's avatar
unknown committed
3711
	srv_io_thread_op_info[segment] = (char *) "complete io for fil node";
3712

3713 3714
	mutex_enter(&(system->mutex));

3715
	fil_node_complete_io(fil_node, fil_system, type);
3716 3717 3718 3719 3720 3721

	mutex_exit(&(system->mutex));

	ut_ad(fil_validate());

	/* Do the i/o handling */
unknown's avatar
unknown committed
3722 3723 3724 3725
	/* IMPORTANT: since i/o handling for reads will read also the insert
	buffer in tablespace 0, you have to be very careful not to introduce
	deadlocks in the i/o system. We keep tablespace 0 data files always
	open, and use a special i/o thread to serve insert buffer requests. */
3726 3727

	if (buf_pool_is_block(message)) {
unknown's avatar
unknown committed
3728 3729
		srv_io_thread_op_info[segment] =
		  (char *) "complete io for buf page";
3730 3731
		buf_page_io_complete(message);
	} else {
unknown's avatar
unknown committed
3732
		srv_io_thread_op_info[segment] =(char *) "complete io for log";
3733 3734 3735 3736 3737
		log_io_complete(message);
	}
}

/**************************************************************************
unknown's avatar
unknown committed
3738 3739
Flushes to disk possible writes cached by the OS. If the space does not exist
or is being dropped, does not do anything. */
3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750

void
fil_flush(
/*======*/
	ulint	space_id)	/* in: file space id (this can be a group of
				log files or a tablespace of the database) */
{
	fil_system_t*	system	= fil_system;
	fil_space_t*	space;
	fil_node_t*	node;
	os_file_t	file;
unknown's avatar
unknown committed
3751
	ib_longlong	old_mod_counter;
3752 3753 3754 3755

	mutex_enter(&(system->mutex));
	
	HASH_SEARCH(hash, system->spaces, space_id, space,
unknown's avatar
unknown committed
3756 3757 3758 3759 3760 3761
							space->id == space_id);
	if (!space || space->is_being_deleted) {
		mutex_exit(&(system->mutex));

		return;
	}
3762

unknown's avatar
unknown committed
3763 3764
	space->n_pending_flushes++;	/* prevent dropping of the space while
					we are flushing */
3765 3766 3767
	node = UT_LIST_GET_FIRST(space->chain);

	while (node) {
unknown's avatar
unknown committed
3768 3769 3770 3771 3772 3773
		if (node->modification_counter > node->flush_counter) {
			ut_a(node->open);

			/* We want to flush the changes at least up to
			old_mod_counter */
			old_mod_counter = node->modification_counter;
3774

3775 3776 3777 3778 3779
			if (space->purpose == FIL_TABLESPACE) {
				fil_n_pending_tablespace_flushes++;
			} else {
				fil_n_pending_log_flushes++;
			}
unknown's avatar
unknown committed
3780 3781
#ifdef __WIN__
			if (node->is_raw_disk) {
3782

unknown's avatar
unknown committed
3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802
				goto skip_flush;
			}
#endif
retry:			
			if (node->n_pending_flushes > 0) {
				/* We want to avoid calling os_file_flush() on
				the file twice at the same time, because we do
				not know what bugs OS's may contain in file
				i/o; sleep for a while */

				mutex_exit(&(system->mutex));

				os_thread_sleep(20000);

				mutex_enter(&(system->mutex));

				if (node->flush_counter >= old_mod_counter) {

					goto skip_flush;
				}
3803

unknown's avatar
unknown committed
3804 3805 3806 3807 3808 3809 3810 3811
				goto retry;
			}

			ut_a(node->open);
			file = node->handle;
			node->n_pending_flushes++;

			mutex_exit(&(system->mutex));
3812 3813

			/* printf("Flushing to file %s\n", node->name); */
unknown's avatar
unknown committed
3814 3815
			os_file_flush(file);		

3816
			mutex_enter(&(system->mutex));
3817

unknown's avatar
unknown committed
3818 3819 3820 3821 3822 3823
			node->n_pending_flushes--;
skip_flush:
			if (node->flush_counter < old_mod_counter) {
				node->flush_counter = old_mod_counter;
			}

3824 3825 3826 3827 3828
			if (space->purpose == FIL_TABLESPACE) {
				fil_n_pending_tablespace_flushes--;
			} else {
				fil_n_pending_log_flushes--;
			}
3829 3830 3831 3832 3833
		}

		node = UT_LIST_GET_NEXT(chain, node);
	}		

unknown's avatar
unknown committed
3834 3835
	space->n_pending_flushes--;

3836 3837 3838 3839
	mutex_exit(&(system->mutex));
}

/**************************************************************************
unknown's avatar
unknown committed
3840
Flushes to disk the writes in file spaces of the given type possibly cached by
3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856
the OS. */

void
fil_flush_file_spaces(
/*==================*/
	ulint	purpose)	/* in: FIL_TABLESPACE, FIL_LOG */
{
	fil_system_t*	system	= fil_system;
	fil_space_t*	space;

	mutex_enter(&(system->mutex));

	space = UT_LIST_GET_FIRST(system->space_list);

	while (space) {
		if (space->purpose == purpose) {
unknown's avatar
unknown committed
3857 3858 3859
			space->n_pending_flushes++; /* prevent dropping of the
						    space while we are
						    flushing */
3860 3861 3862 3863 3864 3865
			mutex_exit(&(system->mutex));

			fil_flush(space->id);

			mutex_enter(&(system->mutex));

unknown's avatar
unknown committed
3866 3867
			space->n_pending_flushes--;
		}
3868 3869 3870 3871 3872 3873 3874
		space = UT_LIST_GET_NEXT(space_list, space);
	}
	
	mutex_exit(&(system->mutex));
}

/**********************************************************************
unknown's avatar
unknown committed
3875
Checks the consistency of the tablespace cache. */
3876 3877 3878 3879 3880 3881

ibool
fil_validate(void)
/*==============*/
			/* out: TRUE if ok */
{	
unknown's avatar
unknown committed
3882
	fil_system_t*	system		= fil_system;
3883 3884
	fil_space_t*	space;
	fil_node_t*	fil_node;
unknown's avatar
unknown committed
3885
	ulint		n_open		= 0;
3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905
	ulint		i;
	
	mutex_enter(&(system->mutex));

	/* Look for spaces in the hash table */

	for (i = 0; i < hash_get_n_cells(system->spaces); i++) {

		space = HASH_GET_FIRST(system->spaces, i);
	
		while (space != NULL) {
			UT_LIST_VALIDATE(chain, fil_node_t, space->chain); 

			fil_node = UT_LIST_GET_FIRST(space->chain);

			while (fil_node != NULL) {
				if (fil_node->n_pending > 0) {
					ut_a(fil_node->open);
				}

unknown's avatar
unknown committed
3906 3907 3908
				if (fil_node->open) {
					n_open++;
				}
3909 3910 3911 3912 3913 3914
				fil_node = UT_LIST_GET_NEXT(chain, fil_node);
			}
			space = HASH_GET_NEXT(hash, space);
		}
	}

unknown's avatar
unknown committed
3915
	ut_a(system->n_open == n_open);
3916 3917 3918 3919 3920 3921 3922 3923

	UT_LIST_VALIDATE(LRU, fil_node_t, system->LRU);

	fil_node = UT_LIST_GET_FIRST(system->LRU);

	while (fil_node != NULL) {
		ut_a(fil_node->n_pending == 0);
		ut_a(fil_node->open);
unknown's avatar
unknown committed
3924 3925
		ut_a(fil_node->space->purpose == FIL_TABLESPACE);
		ut_a(fil_node->space->id != 0);
3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992

		fil_node = UT_LIST_GET_NEXT(LRU, fil_node);
	}
	
	mutex_exit(&(system->mutex));

	return(TRUE);
}

/************************************************************************
Returns TRUE if file address is undefined. */
ibool
fil_addr_is_null(
/*=============*/
				/* out: TRUE if undefined */
	fil_addr_t	addr)	/* in: address */
{
	if (addr.page == FIL_NULL) {

		return(TRUE);
	}

	return(FALSE);
}

/************************************************************************
Accessor functions for a file page */

ulint
fil_page_get_prev(byte*	page)
{
	return(mach_read_from_4(page + FIL_PAGE_PREV));
}

ulint
fil_page_get_next(byte*	page)
{
	return(mach_read_from_4(page + FIL_PAGE_NEXT));
}

/*************************************************************************
Sets the file page type. */

void
fil_page_set_type(
/*==============*/
	byte* 	page,	/* in: file page */
	ulint	type)	/* in: type */
{
	ut_ad(page);

	mach_write_to_2(page + FIL_PAGE_TYPE, type);
}	

/*************************************************************************
Gets the file page type. */

ulint
fil_page_get_type(
/*==============*/
			/* out: type; NOTE that if the type has not been
			written to page, the return value not defined */
	byte* 	page)	/* in: file page */
{
	ut_ad(page);

	return(mach_read_from_2(page + FIL_PAGE_TYPE));
unknown's avatar
unknown committed
3993
}