jfs_txnmgr.c 75.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
/*
 *   Copyright (C) International Business Machines Corp., 2000-2005
 *   Portions Copyright (C) Christoph Hellwig, 2001-2002
 *
 *   This program is free software;  you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation; either version 2 of the License, or 
 *   (at your option) any later version.
 * 
 *   This program is distributed in the hope that it will be useful,
 *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
 *   the GNU General Public License for more details.
 *
 *   You should have received a copy of the GNU General Public License
 *   along with this program;  if not, write to the Free Software 
 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */

/*
 *      jfs_txnmgr.c: transaction manager
 *
 * notes:
 * transaction starts with txBegin() and ends with txCommit()
 * or txAbort().
 *
 * tlock is acquired at the time of update;
 * (obviate scan at commit time for xtree and dtree)
 * tlock and mp points to each other;
 * (no hashlist for mp -> tlock).
 *
 * special cases:
 * tlock on in-memory inode:
 * in-place tlock in the in-memory inode itself;
 * converted to page lock by iWrite() at commit time.
 *
 * tlock during write()/mmap() under anonymous transaction (tid = 0):
 * transferred (?) to transaction at commit time.
 *
 * use the page itself to update allocation maps
 * (obviate intermediate replication of allocation/deallocation data)
 * hold on to mp+lock thru update of maps
 */

#include <linux/fs.h>
#include <linux/vmalloc.h>
#include <linux/smp_lock.h>
#include <linux/completion.h>
#include <linux/suspend.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include "jfs_incore.h"
53
#include "jfs_inode.h"
L
Linus Torvalds 已提交
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 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217
#include "jfs_filsys.h"
#include "jfs_metapage.h"
#include "jfs_dinode.h"
#include "jfs_imap.h"
#include "jfs_dmap.h"
#include "jfs_superblock.h"
#include "jfs_debug.h"

/*
 *      transaction management structures
 */
static struct {
	int freetid;		/* index of a free tid structure */
	int freelock;		/* index first free lock word */
	wait_queue_head_t freewait;	/* eventlist of free tblock */
	wait_queue_head_t freelockwait;	/* eventlist of free tlock */
	wait_queue_head_t lowlockwait;	/* eventlist of ample tlocks */
	int tlocksInUse;	/* Number of tlocks in use */
	spinlock_t LazyLock;	/* synchronize sync_queue & unlock_queue */
/*	struct tblock *sync_queue; * Transactions waiting for data sync */
	struct list_head unlock_queue;	/* Txns waiting to be released */
	struct list_head anon_list;	/* inodes having anonymous txns */
	struct list_head anon_list2;	/* inodes having anonymous txns
					   that couldn't be sync'ed */
} TxAnchor;

int jfs_tlocks_low;		/* Indicates low number of available tlocks */

#ifdef CONFIG_JFS_STATISTICS
static struct {
	uint txBegin;
	uint txBegin_barrier;
	uint txBegin_lockslow;
	uint txBegin_freetid;
	uint txBeginAnon;
	uint txBeginAnon_barrier;
	uint txBeginAnon_lockslow;
	uint txLockAlloc;
	uint txLockAlloc_freelock;
} TxStat;
#endif

static int nTxBlock = -1;	/* number of transaction blocks */
module_param(nTxBlock, int, 0);
MODULE_PARM_DESC(nTxBlock,
		 "Number of transaction blocks (max:65536)");

static int nTxLock = -1;	/* number of transaction locks */
module_param(nTxLock, int, 0);
MODULE_PARM_DESC(nTxLock,
		 "Number of transaction locks (max:65536)");

struct tblock *TxBlock;	        /* transaction block table */
static int TxLockLWM;		/* Low water mark for number of txLocks used */
static int TxLockHWM;		/* High water mark for number of txLocks used */
static int TxLockVHWM;		/* Very High water mark */
struct tlock *TxLock;           /* transaction lock table */

/*
 *      transaction management lock
 */
static DEFINE_SPINLOCK(jfsTxnLock);

#define TXN_LOCK()              spin_lock(&jfsTxnLock)
#define TXN_UNLOCK()            spin_unlock(&jfsTxnLock)

#define LAZY_LOCK_INIT()	spin_lock_init(&TxAnchor.LazyLock);
#define LAZY_LOCK(flags)	spin_lock_irqsave(&TxAnchor.LazyLock, flags)
#define LAZY_UNLOCK(flags) spin_unlock_irqrestore(&TxAnchor.LazyLock, flags)

DECLARE_WAIT_QUEUE_HEAD(jfs_sync_thread_wait);
DECLARE_WAIT_QUEUE_HEAD(jfs_commit_thread_wait);
static int jfs_commit_thread_waking;

/*
 * Retry logic exist outside these macros to protect from spurrious wakeups.
 */
static inline void TXN_SLEEP_DROP_LOCK(wait_queue_head_t * event)
{
	DECLARE_WAITQUEUE(wait, current);

	add_wait_queue(event, &wait);
	set_current_state(TASK_UNINTERRUPTIBLE);
	TXN_UNLOCK();
	schedule();
	current->state = TASK_RUNNING;
	remove_wait_queue(event, &wait);
}

#define TXN_SLEEP(event)\
{\
	TXN_SLEEP_DROP_LOCK(event);\
	TXN_LOCK();\
}

#define TXN_WAKEUP(event) wake_up_all(event)

/*
 *      statistics
 */
static struct {
	tid_t maxtid;		/* 4: biggest tid ever used */
	lid_t maxlid;		/* 4: biggest lid ever used */
	int ntid;		/* 4: # of transactions performed */
	int nlid;		/* 4: # of tlocks acquired */
	int waitlock;		/* 4: # of tlock wait */
} stattx;

/*
 * forward references
 */
static int diLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
		struct tlock * tlck, struct commit * cd);
static int dataLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
		struct tlock * tlck);
static void dtLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
		struct tlock * tlck);
static void mapLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
		struct tlock * tlck);
static void txAllocPMap(struct inode *ip, struct maplock * maplock,
		struct tblock * tblk);
static void txForce(struct tblock * tblk);
static int txLog(struct jfs_log * log, struct tblock * tblk,
		struct commit * cd);
static void txUpdateMap(struct tblock * tblk);
static void txRelease(struct tblock * tblk);
static void xtLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	   struct tlock * tlck);
static void LogSyncRelease(struct metapage * mp);

/*
 *              transaction block/lock management
 *              ---------------------------------
 */

/*
 * Get a transaction lock from the free list.  If the number in use is
 * greater than the high water mark, wake up the sync daemon.  This should
 * free some anonymous transaction locks.  (TXN_LOCK must be held.)
 */
static lid_t txLockAlloc(void)
{
	lid_t lid;

	INCREMENT(TxStat.txLockAlloc);
	if (!TxAnchor.freelock) {
		INCREMENT(TxStat.txLockAlloc_freelock);
	}

	while (!(lid = TxAnchor.freelock))
		TXN_SLEEP(&TxAnchor.freelockwait);
	TxAnchor.freelock = TxLock[lid].next;
	HIGHWATERMARK(stattx.maxlid, lid);
	if ((++TxAnchor.tlocksInUse > TxLockHWM) && (jfs_tlocks_low == 0)) {
		jfs_info("txLockAlloc tlocks low");
		jfs_tlocks_low = 1;
		wake_up(&jfs_sync_thread_wait);
	}

	return lid;
}

static void txLockFree(lid_t lid)
{
218
	TxLock[lid].tid = 0;
L
Linus Torvalds 已提交
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561
	TxLock[lid].next = TxAnchor.freelock;
	TxAnchor.freelock = lid;
	TxAnchor.tlocksInUse--;
	if (jfs_tlocks_low && (TxAnchor.tlocksInUse < TxLockLWM)) {
		jfs_info("txLockFree jfs_tlocks_low no more");
		jfs_tlocks_low = 0;
		TXN_WAKEUP(&TxAnchor.lowlockwait);
	}
	TXN_WAKEUP(&TxAnchor.freelockwait);
}

/*
 * NAME:        txInit()
 *
 * FUNCTION:    initialize transaction management structures
 *
 * RETURN:
 *
 * serialization: single thread at jfs_init()
 */
int txInit(void)
{
	int k, size;
	struct sysinfo si;

	/* Set defaults for nTxLock and nTxBlock if unset */

	if (nTxLock == -1) {
		if (nTxBlock == -1) {
			/* Base default on memory size */
			si_meminfo(&si);
			if (si.totalram > (256 * 1024)) /* 1 GB */
				nTxLock = 64 * 1024;
			else
				nTxLock = si.totalram >> 2;
		} else if (nTxBlock > (8 * 1024))
			nTxLock = 64 * 1024;
		else
			nTxLock = nTxBlock << 3;
	}
	if (nTxBlock == -1)
		nTxBlock = nTxLock >> 3;

	/* Verify tunable parameters */
	if (nTxBlock < 16)
		nTxBlock = 16;	/* No one should set it this low */
	if (nTxBlock > 65536)
		nTxBlock = 65536;
	if (nTxLock < 256)
		nTxLock = 256;	/* No one should set it this low */
	if (nTxLock > 65536)
		nTxLock = 65536;

	printk(KERN_INFO "JFS: nTxBlock = %d, nTxLock = %d\n",
	       nTxBlock, nTxLock);
	/*
	 * initialize transaction block (tblock) table
	 *
	 * transaction id (tid) = tblock index
	 * tid = 0 is reserved.
	 */
	TxLockLWM = (nTxLock * 4) / 10;
	TxLockHWM = (nTxLock * 7) / 10;
	TxLockVHWM = (nTxLock * 8) / 10;

	size = sizeof(struct tblock) * nTxBlock;
	TxBlock = (struct tblock *) vmalloc(size);
	if (TxBlock == NULL)
		return -ENOMEM;

	for (k = 1; k < nTxBlock - 1; k++) {
		TxBlock[k].next = k + 1;
		init_waitqueue_head(&TxBlock[k].gcwait);
		init_waitqueue_head(&TxBlock[k].waitor);
	}
	TxBlock[k].next = 0;
	init_waitqueue_head(&TxBlock[k].gcwait);
	init_waitqueue_head(&TxBlock[k].waitor);

	TxAnchor.freetid = 1;
	init_waitqueue_head(&TxAnchor.freewait);

	stattx.maxtid = 1;	/* statistics */

	/*
	 * initialize transaction lock (tlock) table
	 *
	 * transaction lock id = tlock index
	 * tlock id = 0 is reserved.
	 */
	size = sizeof(struct tlock) * nTxLock;
	TxLock = (struct tlock *) vmalloc(size);
	if (TxLock == NULL) {
		vfree(TxBlock);
		return -ENOMEM;
	}

	/* initialize tlock table */
	for (k = 1; k < nTxLock - 1; k++)
		TxLock[k].next = k + 1;
	TxLock[k].next = 0;
	init_waitqueue_head(&TxAnchor.freelockwait);
	init_waitqueue_head(&TxAnchor.lowlockwait);

	TxAnchor.freelock = 1;
	TxAnchor.tlocksInUse = 0;
	INIT_LIST_HEAD(&TxAnchor.anon_list);
	INIT_LIST_HEAD(&TxAnchor.anon_list2);

	LAZY_LOCK_INIT();
	INIT_LIST_HEAD(&TxAnchor.unlock_queue);

	stattx.maxlid = 1;	/* statistics */

	return 0;
}

/*
 * NAME:        txExit()
 *
 * FUNCTION:    clean up when module is unloaded
 */
void txExit(void)
{
	vfree(TxLock);
	TxLock = NULL;
	vfree(TxBlock);
	TxBlock = NULL;
}

/*
 * NAME:        txBegin()
 *
 * FUNCTION:    start a transaction.
 *
 * PARAMETER:   sb	- superblock
 *              flag	- force for nested tx;
 *
 * RETURN:	tid	- transaction id
 *
 * note: flag force allows to start tx for nested tx
 * to prevent deadlock on logsync barrier;
 */
tid_t txBegin(struct super_block *sb, int flag)
{
	tid_t t;
	struct tblock *tblk;
	struct jfs_log *log;

	jfs_info("txBegin: flag = 0x%x", flag);
	log = JFS_SBI(sb)->log;

	TXN_LOCK();

	INCREMENT(TxStat.txBegin);

      retry:
	if (!(flag & COMMIT_FORCE)) {
		/*
		 * synchronize with logsync barrier
		 */
		if (test_bit(log_SYNCBARRIER, &log->flag) ||
		    test_bit(log_QUIESCE, &log->flag)) {
			INCREMENT(TxStat.txBegin_barrier);
			TXN_SLEEP(&log->syncwait);
			goto retry;
		}
	}
	if (flag == 0) {
		/*
		 * Don't begin transaction if we're getting starved for tlocks
		 * unless COMMIT_FORCE or COMMIT_INODE (which may ultimately
		 * free tlocks)
		 */
		if (TxAnchor.tlocksInUse > TxLockVHWM) {
			INCREMENT(TxStat.txBegin_lockslow);
			TXN_SLEEP(&TxAnchor.lowlockwait);
			goto retry;
		}
	}

	/*
	 * allocate transaction id/block
	 */
	if ((t = TxAnchor.freetid) == 0) {
		jfs_info("txBegin: waiting for free tid");
		INCREMENT(TxStat.txBegin_freetid);
		TXN_SLEEP(&TxAnchor.freewait);
		goto retry;
	}

	tblk = tid_to_tblock(t);

	if ((tblk->next == 0) && !(flag & COMMIT_FORCE)) {
		/* Don't let a non-forced transaction take the last tblk */
		jfs_info("txBegin: waiting for free tid");
		INCREMENT(TxStat.txBegin_freetid);
		TXN_SLEEP(&TxAnchor.freewait);
		goto retry;
	}

	TxAnchor.freetid = tblk->next;

	/*
	 * initialize transaction
	 */

	/*
	 * We can't zero the whole thing or we screw up another thread being
	 * awakened after sleeping on tblk->waitor
	 *
	 * memset(tblk, 0, sizeof(struct tblock));
	 */
	tblk->next = tblk->last = tblk->xflag = tblk->flag = tblk->lsn = 0;

	tblk->sb = sb;
	++log->logtid;
	tblk->logtid = log->logtid;

	++log->active;

	HIGHWATERMARK(stattx.maxtid, t);	/* statistics */
	INCREMENT(stattx.ntid);	/* statistics */

	TXN_UNLOCK();

	jfs_info("txBegin: returning tid = %d", t);

	return t;
}

/*
 * NAME:        txBeginAnon()
 *
 * FUNCTION:    start an anonymous transaction.
 *		Blocks if logsync or available tlocks are low to prevent
 *		anonymous tlocks from depleting supply.
 *
 * PARAMETER:   sb	- superblock
 *
 * RETURN:	none
 */
void txBeginAnon(struct super_block *sb)
{
	struct jfs_log *log;

	log = JFS_SBI(sb)->log;

	TXN_LOCK();
	INCREMENT(TxStat.txBeginAnon);

      retry:
	/*
	 * synchronize with logsync barrier
	 */
	if (test_bit(log_SYNCBARRIER, &log->flag) ||
	    test_bit(log_QUIESCE, &log->flag)) {
		INCREMENT(TxStat.txBeginAnon_barrier);
		TXN_SLEEP(&log->syncwait);
		goto retry;
	}

	/*
	 * Don't begin transaction if we're getting starved for tlocks
	 */
	if (TxAnchor.tlocksInUse > TxLockVHWM) {
		INCREMENT(TxStat.txBeginAnon_lockslow);
		TXN_SLEEP(&TxAnchor.lowlockwait);
		goto retry;
	}
	TXN_UNLOCK();
}

/*
 *      txEnd()
 *
 * function: free specified transaction block.
 *
 *      logsync barrier processing:
 *
 * serialization:
 */
void txEnd(tid_t tid)
{
	struct tblock *tblk = tid_to_tblock(tid);
	struct jfs_log *log;

	jfs_info("txEnd: tid = %d", tid);
	TXN_LOCK();

	/*
	 * wakeup transactions waiting on the page locked
	 * by the current transaction
	 */
	TXN_WAKEUP(&tblk->waitor);

	log = JFS_SBI(tblk->sb)->log;

	/*
	 * Lazy commit thread can't free this guy until we mark it UNLOCKED,
	 * otherwise, we would be left with a transaction that may have been
	 * reused.
	 *
	 * Lazy commit thread will turn off tblkGC_LAZY before calling this
	 * routine.
	 */
	if (tblk->flag & tblkGC_LAZY) {
		jfs_info("txEnd called w/lazy tid: %d, tblk = 0x%p", tid, tblk);
		TXN_UNLOCK();

		spin_lock_irq(&log->gclock);	// LOGGC_LOCK
		tblk->flag |= tblkGC_UNLOCKED;
		spin_unlock_irq(&log->gclock);	// LOGGC_UNLOCK
		return;
	}

	jfs_info("txEnd: tid: %d, tblk = 0x%p", tid, tblk);

	assert(tblk->next == 0);

	/*
	 * insert tblock back on freelist
	 */
	tblk->next = TxAnchor.freetid;
	TxAnchor.freetid = tid;

	/*
	 * mark the tblock not active
	 */
	if (--log->active == 0) {
		clear_bit(log_FLUSH, &log->flag);

		/*
		 * synchronize with logsync barrier
		 */
		if (test_bit(log_SYNCBARRIER, &log->flag)) {
			jfs_info("log barrier off: 0x%x", log->lsn);

			/* enable new transactions start */
			clear_bit(log_SYNCBARRIER, &log->flag);

			/* wakeup all waitors for logsync barrier */
			TXN_WAKEUP(&log->syncwait);
562 563 564 565 566 567 568

			TXN_UNLOCK();

			/* forward log syncpt */
			jfs_syncpt(log);

			goto wakeup;
L
Linus Torvalds 已提交
569 570 571
		}
	}

572 573
	TXN_UNLOCK();
wakeup:
L
Linus Torvalds 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
	/*
	 * wakeup all waitors for a free tblock
	 */
	TXN_WAKEUP(&TxAnchor.freewait);
}

/*
 *      txLock()
 *
 * function: acquire a transaction lock on the specified <mp>
 *
 * parameter:
 *
 * return:      transaction lock id
 *
 * serialization:
 */
struct tlock *txLock(tid_t tid, struct inode *ip, struct metapage * mp,
		     int type)
{
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	int dir_xtree = 0;
	lid_t lid;
	tid_t xtid;
	struct tlock *tlck;
	struct xtlock *xtlck;
	struct linelock *linelock;
	xtpage_t *p;
	struct tblock *tblk;

	TXN_LOCK();

	if (S_ISDIR(ip->i_mode) && (type & tlckXTREE) &&
	    !(mp->xflag & COMMIT_PAGE)) {
		/*
		 * Directory inode is special.  It can have both an xtree tlock
		 * and a dtree tlock associated with it.
		 */
		dir_xtree = 1;
		lid = jfs_ip->xtlid;
	} else
		lid = mp->lid;

	/* is page not locked by a transaction ? */
	if (lid == 0)
		goto allocateLock;

	jfs_info("txLock: tid:%d ip:0x%p mp:0x%p lid:%d", tid, ip, mp, lid);

	/* is page locked by the requester transaction ? */
	tlck = lid_to_tlock(lid);
625 626
	if ((xtid = tlck->tid) == tid) {
		TXN_UNLOCK();
L
Linus Torvalds 已提交
627
		goto grantLock;
628
	}
L
Linus Torvalds 已提交
629 630 631 632 633 634 635 636 637 638 639 640 641 642

	/*
	 * is page locked by anonymous transaction/lock ?
	 *
	 * (page update without transaction (i.e., file write) is
	 * locked under anonymous transaction tid = 0:
	 * anonymous tlocks maintained on anonymous tlock list of
	 * the inode of the page and available to all anonymous
	 * transactions until txCommit() time at which point
	 * they are transferred to the transaction tlock list of
	 * the commiting transaction of the inode)
	 */
	if (xtid == 0) {
		tlck->tid = tid;
643
		TXN_UNLOCK();
L
Linus Torvalds 已提交
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
		tblk = tid_to_tblock(tid);
		/*
		 * The order of the tlocks in the transaction is important
		 * (during truncate, child xtree pages must be freed before
		 * parent's tlocks change the working map).
		 * Take tlock off anonymous list and add to tail of
		 * transaction list
		 *
		 * Note:  We really need to get rid of the tid & lid and
		 * use list_head's.  This code is getting UGLY!
		 */
		if (jfs_ip->atlhead == lid) {
			if (jfs_ip->atltail == lid) {
				/* only anonymous txn.
				 * Remove from anon_list
				 */
				list_del_init(&jfs_ip->anon_inode_list);
			}
			jfs_ip->atlhead = tlck->next;
		} else {
			lid_t last;
			for (last = jfs_ip->atlhead;
			     lid_to_tlock(last)->next != lid;
			     last = lid_to_tlock(last)->next) {
				assert(last);
			}
			lid_to_tlock(last)->next = tlck->next;
			if (jfs_ip->atltail == lid)
				jfs_ip->atltail = last;
		}

		/* insert the tlock at tail of transaction tlock list */

		if (tblk->next)
			lid_to_tlock(tblk->last)->next = lid;
		else
			tblk->next = lid;
		tlck->next = 0;
		tblk->last = lid;

		goto grantLock;
	}

	goto waitLock;

	/*
	 * allocate a tlock
	 */
      allocateLock:
	lid = txLockAlloc();
	tlck = lid_to_tlock(lid);

	/*
	 * initialize tlock
	 */
	tlck->tid = tid;

701 702
	TXN_UNLOCK();

L
Linus Torvalds 已提交
703 704 705 706 707 708
	/* mark tlock for meta-data page */
	if (mp->xflag & COMMIT_PAGE) {

		tlck->flag = tlckPAGELOCK;

		/* mark the page dirty and nohomeok */
709
		metapage_nohomeok(mp);
L
Linus Torvalds 已提交
710 711

		jfs_info("locking mp = 0x%p, nohomeok = %d tid = %d tlck = 0x%p",
712
			 mp, mp->nohomeok, tid, tlck);
L
Linus Torvalds 已提交
713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757

		/* if anonymous transaction, and buffer is on the group
		 * commit synclist, mark inode to show this.  This will
		 * prevent the buffer from being marked nohomeok for too
		 * long a time.
		 */
		if ((tid == 0) && mp->lsn)
			set_cflag(COMMIT_Synclist, ip);
	}
	/* mark tlock for in-memory inode */
	else
		tlck->flag = tlckINODELOCK;

	tlck->type = 0;

	/* bind the tlock and the page */
	tlck->ip = ip;
	tlck->mp = mp;
	if (dir_xtree)
		jfs_ip->xtlid = lid;
	else
		mp->lid = lid;

	/*
	 * enqueue transaction lock to transaction/inode
	 */
	/* insert the tlock at tail of transaction tlock list */
	if (tid) {
		tblk = tid_to_tblock(tid);
		if (tblk->next)
			lid_to_tlock(tblk->last)->next = lid;
		else
			tblk->next = lid;
		tlck->next = 0;
		tblk->last = lid;
	}
	/* anonymous transaction:
	 * insert the tlock at head of inode anonymous tlock list
	 */
	else {
		tlck->next = jfs_ip->atlhead;
		jfs_ip->atlhead = lid;
		if (tlck->next == 0) {
			/* This inode's first anonymous transaction */
			jfs_ip->atltail = lid;
758
			TXN_LOCK();
L
Linus Torvalds 已提交
759 760
			list_add_tail(&jfs_ip->anon_inode_list,
				      &TxAnchor.anon_list);
761
			TXN_UNLOCK();
L
Linus Torvalds 已提交
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
		}
	}

	/* initialize type dependent area for linelock */
	linelock = (struct linelock *) & tlck->lock;
	linelock->next = 0;
	linelock->flag = tlckLINELOCK;
	linelock->maxcnt = TLOCKSHORT;
	linelock->index = 0;

	switch (type & tlckTYPE) {
	case tlckDTREE:
		linelock->l2linesize = L2DTSLOTSIZE;
		break;

	case tlckXTREE:
		linelock->l2linesize = L2XTSLOTSIZE;

		xtlck = (struct xtlock *) linelock;
		xtlck->header.offset = 0;
		xtlck->header.length = 2;

		if (type & tlckNEW) {
			xtlck->lwm.offset = XTENTRYSTART;
		} else {
			if (mp->xflag & COMMIT_PAGE)
				p = (xtpage_t *) mp->data;
			else
				p = &jfs_ip->i_xtroot;
			xtlck->lwm.offset =
			    le16_to_cpu(p->header.nextindex);
		}
		xtlck->lwm.length = 0;	/* ! */
		xtlck->twm.offset = 0;
		xtlck->hwm.offset = 0;

		xtlck->index = 2;
		break;

	case tlckINODE:
		linelock->l2linesize = L2INODESLOTSIZE;
		break;

	case tlckDATA:
		linelock->l2linesize = L2DATASLOTSIZE;
		break;

	default:
		jfs_err("UFO tlock:0x%p", tlck);
	}

	/*
	 * update tlock vector
	 */
      grantLock:
	tlck->type |= type;

	return tlck;

	/*
	 * page is being locked by another transaction:
	 */
      waitLock:
	/* Only locks on ipimap or ipaimap should reach here */
	/* assert(jfs_ip->fileset == AGGREGATE_I); */
	if (jfs_ip->fileset != AGGREGATE_I) {
		jfs_err("txLock: trying to lock locked page!");
		dump_mem("ip", ip, sizeof(struct inode));
		dump_mem("mp", mp, sizeof(struct metapage));
		dump_mem("Locker's tblk", tid_to_tblock(tid),
			 sizeof(struct tblock));
		dump_mem("Tlock", tlck, sizeof(struct tlock));
		BUG();
	}
	INCREMENT(stattx.waitlock);	/* statistics */
837
	TXN_UNLOCK();
L
Linus Torvalds 已提交
838
	release_metapage(mp);
839 840
	TXN_LOCK();
	xtid = tlck->tid;	/* reaquire after dropping TXN_LOCK */
L
Linus Torvalds 已提交
841 842 843

	jfs_info("txLock: in waitLock, tid = %d, xtid = %d, lid = %d",
		 tid, xtid, lid);
844 845 846 847 848 849

	/* Recheck everything since dropping TXN_LOCK */
	if (xtid && (tlck->mp == mp) && (mp->lid == lid))
		TXN_SLEEP_DROP_LOCK(&tid_to_tblock(xtid)->waitor);
	else
		TXN_UNLOCK();
L
Linus Torvalds 已提交
850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907
	jfs_info("txLock: awakened     tid = %d, lid = %d", tid, lid);

	return NULL;
}

/*
 * NAME:        txRelease()
 *
 * FUNCTION:    Release buffers associated with transaction locks, but don't
 *		mark homeok yet.  The allows other transactions to modify
 *		buffers, but won't let them go to disk until commit record
 *		actually gets written.
 *
 * PARAMETER:
 *              tblk    -
 *
 * RETURN:      Errors from subroutines.
 */
static void txRelease(struct tblock * tblk)
{
	struct metapage *mp;
	lid_t lid;
	struct tlock *tlck;

	TXN_LOCK();

	for (lid = tblk->next; lid; lid = tlck->next) {
		tlck = lid_to_tlock(lid);
		if ((mp = tlck->mp) != NULL &&
		    (tlck->type & tlckBTROOT) == 0) {
			assert(mp->xflag & COMMIT_PAGE);
			mp->lid = 0;
		}
	}

	/*
	 * wakeup transactions waiting on a page locked
	 * by the current transaction
	 */
	TXN_WAKEUP(&tblk->waitor);

	TXN_UNLOCK();
}

/*
 * NAME:        txUnlock()
 *
 * FUNCTION:    Initiates pageout of pages modified by tid in journalled
 *              objects and frees their lockwords.
 */
static void txUnlock(struct tblock * tblk)
{
	struct tlock *tlck;
	struct linelock *linelock;
	lid_t lid, next, llid, k;
	struct metapage *mp;
	struct jfs_log *log;
	int difft, diffp;
908
	unsigned long flags;
L
Linus Torvalds 已提交
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928

	jfs_info("txUnlock: tblk = 0x%p", tblk);
	log = JFS_SBI(tblk->sb)->log;

	/*
	 * mark page under tlock homeok (its log has been written):
	 */
	for (lid = tblk->next; lid; lid = next) {
		tlck = lid_to_tlock(lid);
		next = tlck->next;

		jfs_info("unlocking lid = %d, tlck = 0x%p", lid, tlck);

		/* unbind page from tlock */
		if ((mp = tlck->mp) != NULL &&
		    (tlck->type & tlckBTROOT) == 0) {
			assert(mp->xflag & COMMIT_PAGE);

			/* hold buffer
			 */
929
			hold_metapage(mp);
L
Linus Torvalds 已提交
930

931 932
			assert(mp->nohomeok > 0);
			_metapage_homeok(mp);
L
Linus Torvalds 已提交
933 934

			/* inherit younger/larger clsn */
935
			LOGSYNC_LOCK(log, flags);
L
Linus Torvalds 已提交
936 937 938 939 940 941 942
			if (mp->clsn) {
				logdiff(difft, tblk->clsn, log);
				logdiff(diffp, mp->clsn, log);
				if (difft > diffp)
					mp->clsn = tblk->clsn;
			} else
				mp->clsn = tblk->clsn;
943
			LOGSYNC_UNLOCK(log, flags);
L
Linus Torvalds 已提交
944 945 946

			assert(!(tlck->flag & tlckFREEPAGE));

947
			put_metapage(mp);
L
Linus Torvalds 已提交
948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973
		}

		/* insert tlock, and linelock(s) of the tlock if any,
		 * at head of freelist
		 */
		TXN_LOCK();

		llid = ((struct linelock *) & tlck->lock)->next;
		while (llid) {
			linelock = (struct linelock *) lid_to_tlock(llid);
			k = linelock->next;
			txLockFree(llid);
			llid = k;
		}
		txLockFree(lid);

		TXN_UNLOCK();
	}
	tblk->next = tblk->last = 0;

	/*
	 * remove tblock from logsynclist
	 * (allocation map pages inherited lsn of tblk and
	 * has been inserted in logsync list at txUpdateMap())
	 */
	if (tblk->lsn) {
974
		LOGSYNC_LOCK(log, flags);
L
Linus Torvalds 已提交
975 976
		log->count--;
		list_del(&tblk->synclist);
977
		LOGSYNC_UNLOCK(log, flags);
L
Linus Torvalds 已提交
978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 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 1355 1356 1357 1358 1359 1360 1361 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 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
	}
}

/*
 *      txMaplock()
 *
 * function: allocate a transaction lock for freed page/entry;
 *      for freed page, maplock is used as xtlock/dtlock type;
 */
struct tlock *txMaplock(tid_t tid, struct inode *ip, int type)
{
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	lid_t lid;
	struct tblock *tblk;
	struct tlock *tlck;
	struct maplock *maplock;

	TXN_LOCK();

	/*
	 * allocate a tlock
	 */
	lid = txLockAlloc();
	tlck = lid_to_tlock(lid);

	/*
	 * initialize tlock
	 */
	tlck->tid = tid;

	/* bind the tlock and the object */
	tlck->flag = tlckINODELOCK;
	tlck->ip = ip;
	tlck->mp = NULL;

	tlck->type = type;

	/*
	 * enqueue transaction lock to transaction/inode
	 */
	/* insert the tlock at tail of transaction tlock list */
	if (tid) {
		tblk = tid_to_tblock(tid);
		if (tblk->next)
			lid_to_tlock(tblk->last)->next = lid;
		else
			tblk->next = lid;
		tlck->next = 0;
		tblk->last = lid;
	}
	/* anonymous transaction:
	 * insert the tlock at head of inode anonymous tlock list
	 */
	else {
		tlck->next = jfs_ip->atlhead;
		jfs_ip->atlhead = lid;
		if (tlck->next == 0) {
			/* This inode's first anonymous transaction */
			jfs_ip->atltail = lid;
			list_add_tail(&jfs_ip->anon_inode_list,
				      &TxAnchor.anon_list);
		}
	}

	TXN_UNLOCK();

	/* initialize type dependent area for maplock */
	maplock = (struct maplock *) & tlck->lock;
	maplock->next = 0;
	maplock->maxcnt = 0;
	maplock->index = 0;

	return tlck;
}

/*
 *      txLinelock()
 *
 * function: allocate a transaction lock for log vector list
 */
struct linelock *txLinelock(struct linelock * tlock)
{
	lid_t lid;
	struct tlock *tlck;
	struct linelock *linelock;

	TXN_LOCK();

	/* allocate a TxLock structure */
	lid = txLockAlloc();
	tlck = lid_to_tlock(lid);

	TXN_UNLOCK();

	/* initialize linelock */
	linelock = (struct linelock *) tlck;
	linelock->next = 0;
	linelock->flag = tlckLINELOCK;
	linelock->maxcnt = TLOCKLONG;
	linelock->index = 0;

	/* append linelock after tlock */
	linelock->next = tlock->next;
	tlock->next = lid;

	return linelock;
}

/*
 *              transaction commit management
 *              -----------------------------
 */

/*
 * NAME:        txCommit()
 *
 * FUNCTION:    commit the changes to the objects specified in
 *              clist.  For journalled segments only the
 *              changes of the caller are committed, ie by tid.
 *              for non-journalled segments the data are flushed to
 *              disk and then the change to the disk inode and indirect
 *              blocks committed (so blocks newly allocated to the
 *              segment will be made a part of the segment atomically).
 *
 *              all of the segments specified in clist must be in
 *              one file system. no more than 6 segments are needed
 *              to handle all unix svcs.
 *
 *              if the i_nlink field (i.e. disk inode link count)
 *              is zero, and the type of inode is a regular file or
 *              directory, or symbolic link , the inode is truncated
 *              to zero length. the truncation is committed but the
 *              VM resources are unaffected until it is closed (see
 *              iput and iclose).
 *
 * PARAMETER:
 *
 * RETURN:
 *
 * serialization:
 *              on entry the inode lock on each segment is assumed
 *              to be held.
 *
 * i/o error:
 */
int txCommit(tid_t tid,		/* transaction identifier */
	     int nip,		/* number of inodes to commit */
	     struct inode **iplist,	/* list of inode to commit */
	     int flag)
{
	int rc = 0;
	struct commit cd;
	struct jfs_log *log;
	struct tblock *tblk;
	struct lrd *lrd;
	int lsn;
	struct inode *ip;
	struct jfs_inode_info *jfs_ip;
	int k, n;
	ino_t top;
	struct super_block *sb;

	jfs_info("txCommit, tid = %d, flag = %d", tid, flag);
	/* is read-only file system ? */
	if (isReadOnly(iplist[0])) {
		rc = -EROFS;
		goto TheEnd;
	}

	sb = cd.sb = iplist[0]->i_sb;
	cd.tid = tid;

	if (tid == 0)
		tid = txBegin(sb, 0);
	tblk = tid_to_tblock(tid);

	/*
	 * initialize commit structure
	 */
	log = JFS_SBI(sb)->log;
	cd.log = log;

	/* initialize log record descriptor in commit */
	lrd = &cd.lrd;
	lrd->logtid = cpu_to_le32(tblk->logtid);
	lrd->backchain = 0;

	tblk->xflag |= flag;

	if ((flag & (COMMIT_FORCE | COMMIT_SYNC)) == 0)
		tblk->xflag |= COMMIT_LAZY;
	/*
	 *      prepare non-journaled objects for commit
	 *
	 * flush data pages of non-journaled file
	 * to prevent the file getting non-initialized disk blocks
	 * in case of crash.
	 * (new blocks - )
	 */
	cd.iplist = iplist;
	cd.nip = nip;

	/*
	 *      acquire transaction lock on (on-disk) inodes
	 *
	 * update on-disk inode from in-memory inode
	 * acquiring transaction locks for AFTER records
	 * on the on-disk inode of file object
	 *
	 * sort the inodes array by inode number in descending order
	 * to prevent deadlock when acquiring transaction lock
	 * of on-disk inodes on multiple on-disk inode pages by
	 * multiple concurrent transactions
	 */
	for (k = 0; k < cd.nip; k++) {
		top = (cd.iplist[k])->i_ino;
		for (n = k + 1; n < cd.nip; n++) {
			ip = cd.iplist[n];
			if (ip->i_ino > top) {
				top = ip->i_ino;
				cd.iplist[n] = cd.iplist[k];
				cd.iplist[k] = ip;
			}
		}

		ip = cd.iplist[k];
		jfs_ip = JFS_IP(ip);

		/*
		 * BUGBUG - This code has temporarily been removed.  The
		 * intent is to ensure that any file data is written before
		 * the metadata is committed to the journal.  This prevents
		 * uninitialized data from appearing in a file after the
		 * journal has been replayed.  (The uninitialized data
		 * could be sensitive data removed by another user.)
		 *
		 * The problem now is that we are holding the IWRITELOCK
		 * on the inode, and calling filemap_fdatawrite on an
		 * unmapped page will cause a deadlock in jfs_get_block.
		 *
		 * The long term solution is to pare down the use of
		 * IWRITELOCK.  We are currently holding it too long.
		 * We could also be smarter about which data pages need
		 * to be written before the transaction is committed and
		 * when we don't need to worry about it at all.
		 *
		 * if ((!S_ISDIR(ip->i_mode))
		 *    && (tblk->flag & COMMIT_DELETE) == 0) {
		 *	filemap_fdatawrite(ip->i_mapping);
		 *	filemap_fdatawait(ip->i_mapping);
		 * }
		 */

		/*
		 * Mark inode as not dirty.  It will still be on the dirty
		 * inode list, but we'll know not to commit it again unless
		 * it gets marked dirty again
		 */
		clear_cflag(COMMIT_Dirty, ip);

		/* inherit anonymous tlock(s) of inode */
		if (jfs_ip->atlhead) {
			lid_to_tlock(jfs_ip->atltail)->next = tblk->next;
			tblk->next = jfs_ip->atlhead;
			if (!tblk->last)
				tblk->last = jfs_ip->atltail;
			jfs_ip->atlhead = jfs_ip->atltail = 0;
			TXN_LOCK();
			list_del_init(&jfs_ip->anon_inode_list);
			TXN_UNLOCK();
		}

		/*
		 * acquire transaction lock on on-disk inode page
		 * (become first tlock of the tblk's tlock list)
		 */
		if (((rc = diWrite(tid, ip))))
			goto out;
	}

	/*
	 *      write log records from transaction locks
	 *
	 * txUpdateMap() resets XAD_NEW in XAD.
	 */
	if ((rc = txLog(log, tblk, &cd)))
		goto TheEnd;

	/*
	 * Ensure that inode isn't reused before
	 * lazy commit thread finishes processing
	 */
	if (tblk->xflag & COMMIT_DELETE) {
		atomic_inc(&tblk->u.ip->i_count);
		/*
		 * Avoid a rare deadlock
		 *
		 * If the inode is locked, we may be blocked in
		 * jfs_commit_inode.  If so, we don't want the
		 * lazy_commit thread doing the last iput() on the inode
		 * since that may block on the locked inode.  Instead,
		 * commit the transaction synchronously, so the last iput
		 * will be done by the calling thread (or later)
		 */
		if (tblk->u.ip->i_state & I_LOCK)
			tblk->xflag &= ~COMMIT_LAZY;
	}

	ASSERT((!(tblk->xflag & COMMIT_DELETE)) ||
	       ((tblk->u.ip->i_nlink == 0) &&
		!test_cflag(COMMIT_Nolink, tblk->u.ip)));

	/*
	 *      write COMMIT log record
	 */
	lrd->type = cpu_to_le16(LOG_COMMIT);
	lrd->length = 0;
	lsn = lmLog(log, tblk, lrd, NULL);

	lmGroupCommit(log, tblk);

	/*
	 *      - transaction is now committed -
	 */

	/*
	 * force pages in careful update
	 * (imap addressing structure update)
	 */
	if (flag & COMMIT_FORCE)
		txForce(tblk);

	/*
	 *      update allocation map.
	 *
	 * update inode allocation map and inode:
	 * free pager lock on memory object of inode if any.
	 * update  block allocation map.
	 *
	 * txUpdateMap() resets XAD_NEW in XAD.
	 */
	if (tblk->xflag & COMMIT_FORCE)
		txUpdateMap(tblk);

	/*
	 *      free transaction locks and pageout/free pages
	 */
	txRelease(tblk);

	if ((tblk->flag & tblkGC_LAZY) == 0)
		txUnlock(tblk);


	/*
	 *      reset in-memory object state
	 */
	for (k = 0; k < cd.nip; k++) {
		ip = cd.iplist[k];
		jfs_ip = JFS_IP(ip);

		/*
		 * reset in-memory inode state
		 */
		jfs_ip->bxflag = 0;
		jfs_ip->blid = 0;
	}

      out:
	if (rc != 0)
		txAbort(tid, 1);

      TheEnd:
	jfs_info("txCommit: tid = %d, returning %d", tid, rc);
	return rc;
}

/*
 * NAME:        txLog()
 *
 * FUNCTION:    Writes AFTER log records for all lines modified
 *              by tid for segments specified by inodes in comdata.
 *              Code assumes only WRITELOCKS are recorded in lockwords.
 *
 * PARAMETERS:
 *
 * RETURN :
 */
static int txLog(struct jfs_log * log, struct tblock * tblk, struct commit * cd)
{
	int rc = 0;
	struct inode *ip;
	lid_t lid;
	struct tlock *tlck;
	struct lrd *lrd = &cd->lrd;

	/*
	 * write log record(s) for each tlock of transaction,
	 */
	for (lid = tblk->next; lid; lid = tlck->next) {
		tlck = lid_to_tlock(lid);

		tlck->flag |= tlckLOG;

		/* initialize lrd common */
		ip = tlck->ip;
		lrd->aggregate = cpu_to_le32(JFS_SBI(ip->i_sb)->aggregate);
		lrd->log.redopage.fileset = cpu_to_le32(JFS_IP(ip)->fileset);
		lrd->log.redopage.inode = cpu_to_le32(ip->i_ino);

		/* write log record of page from the tlock */
		switch (tlck->type & tlckTYPE) {
		case tlckXTREE:
			xtLog(log, tblk, lrd, tlck);
			break;

		case tlckDTREE:
			dtLog(log, tblk, lrd, tlck);
			break;

		case tlckINODE:
			diLog(log, tblk, lrd, tlck, cd);
			break;

		case tlckMAP:
			mapLog(log, tblk, lrd, tlck);
			break;

		case tlckDATA:
			dataLog(log, tblk, lrd, tlck);
			break;

		default:
			jfs_err("UFO tlock:0x%p", tlck);
		}
	}

	return rc;
}

/*
 *      diLog()
 *
 * function:    log inode tlock and format maplock to update bmap;
 */
static int diLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	  struct tlock * tlck, struct commit * cd)
{
	int rc = 0;
	struct metapage *mp;
	pxd_t *pxd;
	struct pxd_lock *pxdlock;

	mp = tlck->mp;

	/* initialize as REDOPAGE record format */
	lrd->log.redopage.type = cpu_to_le16(LOG_INODE);
	lrd->log.redopage.l2linesize = cpu_to_le16(L2INODESLOTSIZE);

	pxd = &lrd->log.redopage.pxd;

	/*
	 *      inode after image
	 */
	if (tlck->type & tlckENTRY) {
		/* log after-image for logredo(): */
		lrd->type = cpu_to_le16(LOG_REDOPAGE);
		PXDaddress(pxd, mp->index);
		PXDlength(pxd,
			  mp->logical_size >> tblk->sb->s_blocksize_bits);
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;
	} else if (tlck->type & tlckFREE) {
		/*
		 *      free inode extent
		 *
		 * (pages of the freed inode extent have been invalidated and
		 * a maplock for free of the extent has been formatted at
		 * txLock() time);
		 *
		 * the tlock had been acquired on the inode allocation map page
		 * (iag) that specifies the freed extent, even though the map
		 * page is not itself logged, to prevent pageout of the map
		 * page before the log;
		 */

		/* log LOG_NOREDOINOEXT of the freed inode extent for
		 * logredo() to start NoRedoPage filters, and to update
		 * imap and bmap for free of the extent;
		 */
		lrd->type = cpu_to_le16(LOG_NOREDOINOEXT);
		/*
		 * For the LOG_NOREDOINOEXT record, we need
		 * to pass the IAG number and inode extent
		 * index (within that IAG) from which the
		 * the extent being released.  These have been
		 * passed to us in the iplist[1] and iplist[2].
		 */
		lrd->log.noredoinoext.iagnum =
		    cpu_to_le32((u32) (size_t) cd->iplist[1]);
		lrd->log.noredoinoext.inoext_idx =
		    cpu_to_le32((u32) (size_t) cd->iplist[2]);

		pxdlock = (struct pxd_lock *) & tlck->lock;
		*pxd = pxdlock->pxd;
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, NULL));

		/* update bmap */
		tlck->flag |= tlckUPDATEMAP;

		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;
	} else
		jfs_err("diLog: UFO type tlck:0x%p", tlck);
#ifdef  _JFS_WIP
	/*
	 *      alloc/free external EA extent
	 *
	 * a maplock for txUpdateMap() to update bPWMAP for alloc/free
	 * of the extent has been formatted at txLock() time;
	 */
	else {
		assert(tlck->type & tlckEA);

		/* log LOG_UPDATEMAP for logredo() to update bmap for
		 * alloc of new (and free of old) external EA extent;
		 */
		lrd->type = cpu_to_le16(LOG_UPDATEMAP);
		pxdlock = (struct pxd_lock *) & tlck->lock;
		nlock = pxdlock->index;
		for (i = 0; i < nlock; i++, pxdlock++) {
			if (pxdlock->flag & mlckALLOCPXD)
				lrd->log.updatemap.type =
				    cpu_to_le16(LOG_ALLOCPXD);
			else
				lrd->log.updatemap.type =
				    cpu_to_le16(LOG_FREEPXD);
			lrd->log.updatemap.nxd = cpu_to_le16(1);
			lrd->log.updatemap.pxd = pxdlock->pxd;
			lrd->backchain =
			    cpu_to_le32(lmLog(log, tblk, lrd, NULL));
		}

		/* update bmap */
		tlck->flag |= tlckUPDATEMAP;
	}
#endif				/* _JFS_WIP */

	return rc;
}

/*
 *      dataLog()
 *
 * function:    log data tlock
 */
static int dataLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	    struct tlock * tlck)
{
	struct metapage *mp;
	pxd_t *pxd;

	mp = tlck->mp;

	/* initialize as REDOPAGE record format */
	lrd->log.redopage.type = cpu_to_le16(LOG_DATA);
	lrd->log.redopage.l2linesize = cpu_to_le16(L2DATASLOTSIZE);

	pxd = &lrd->log.redopage.pxd;

	/* log after-image for logredo(): */
	lrd->type = cpu_to_le16(LOG_REDOPAGE);

	if (jfs_dirtable_inline(tlck->ip)) {
		/*
		 * The table has been truncated, we've must have deleted
		 * the last entry, so don't bother logging this
		 */
		mp->lid = 0;
1558 1559
		grab_metapage(mp);
		metapage_homeok(mp);
L
Linus Torvalds 已提交
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
		discard_metapage(mp);
		tlck->mp = NULL;
		return 0;
	}

	PXDaddress(pxd, mp->index);
	PXDlength(pxd, mp->logical_size >> tblk->sb->s_blocksize_bits);

	lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

	/* mark page as homeward bound */
	tlck->flag |= tlckWRITEPAGE;

	return 0;
}

/*
 *      dtLog()
 *
 * function:    log dtree tlock and format maplock to update bmap;
 */
static void dtLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	   struct tlock * tlck)
{
	struct metapage *mp;
	struct pxd_lock *pxdlock;
	pxd_t *pxd;

	mp = tlck->mp;

	/* initialize as REDOPAGE/NOREDOPAGE record format */
	lrd->log.redopage.type = cpu_to_le16(LOG_DTREE);
	lrd->log.redopage.l2linesize = cpu_to_le16(L2DTSLOTSIZE);

	pxd = &lrd->log.redopage.pxd;

	if (tlck->type & tlckBTROOT)
		lrd->log.redopage.type |= cpu_to_le16(LOG_BTROOT);

	/*
	 *      page extension via relocation: entry insertion;
	 *      page extension in-place: entry insertion;
	 *      new right page from page split, reinitialized in-line
	 *      root from root page split: entry insertion;
	 */
	if (tlck->type & (tlckNEW | tlckEXTEND)) {
		/* log after-image of the new page for logredo():
		 * mark log (LOG_NEW) for logredo() to initialize
		 * freelist and update bmap for alloc of the new page;
		 */
		lrd->type = cpu_to_le16(LOG_REDOPAGE);
		if (tlck->type & tlckEXTEND)
			lrd->log.redopage.type |= cpu_to_le16(LOG_EXTEND);
		else
			lrd->log.redopage.type |= cpu_to_le16(LOG_NEW);
		PXDaddress(pxd, mp->index);
		PXDlength(pxd,
			  mp->logical_size >> tblk->sb->s_blocksize_bits);
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/* format a maplock for txUpdateMap() to update bPMAP for
		 * alloc of the new page;
		 */
		if (tlck->type & tlckBTROOT)
			return;
		tlck->flag |= tlckUPDATEMAP;
		pxdlock = (struct pxd_lock *) & tlck->lock;
		pxdlock->flag = mlckALLOCPXD;
		pxdlock->pxd = *pxd;

		pxdlock->index = 1;

		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;
		return;
	}

	/*
	 *      entry insertion/deletion,
	 *      sibling page link update (old right page before split);
	 */
	if (tlck->type & (tlckENTRY | tlckRELINK)) {
		/* log after-image for logredo(): */
		lrd->type = cpu_to_le16(LOG_REDOPAGE);
		PXDaddress(pxd, mp->index);
		PXDlength(pxd,
			  mp->logical_size >> tblk->sb->s_blocksize_bits);
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;
		return;
	}

	/*
	 *      page deletion: page has been invalidated
	 *      page relocation: source extent
	 *
	 *      a maplock for free of the page has been formatted
	 *      at txLock() time);
	 */
	if (tlck->type & (tlckFREE | tlckRELOCATE)) {
		/* log LOG_NOREDOPAGE of the deleted page for logredo()
		 * to start NoRedoPage filter and to update bmap for free
		 * of the deletd page
		 */
		lrd->type = cpu_to_le16(LOG_NOREDOPAGE);
		pxdlock = (struct pxd_lock *) & tlck->lock;
		*pxd = pxdlock->pxd;
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, NULL));

		/* a maplock for txUpdateMap() for free of the page
		 * has been formatted at txLock() time;
		 */
		tlck->flag |= tlckUPDATEMAP;
	}
	return;
}

/*
 *      xtLog()
 *
 * function:    log xtree tlock and format maplock to update bmap;
 */
static void xtLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	   struct tlock * tlck)
{
	struct inode *ip;
	struct metapage *mp;
	xtpage_t *p;
	struct xtlock *xtlck;
	struct maplock *maplock;
	struct xdlistlock *xadlock;
	struct pxd_lock *pxdlock;
1694
	pxd_t *page_pxd;
L
Linus Torvalds 已提交
1695 1696 1697 1698 1699 1700 1701 1702 1703
	int next, lwm, hwm;

	ip = tlck->ip;
	mp = tlck->mp;

	/* initialize as REDOPAGE/NOREDOPAGE record format */
	lrd->log.redopage.type = cpu_to_le16(LOG_XTREE);
	lrd->log.redopage.l2linesize = cpu_to_le16(L2XTSLOTSIZE);

1704
	page_pxd = &lrd->log.redopage.pxd;
L
Linus Torvalds 已提交
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733

	if (tlck->type & tlckBTROOT) {
		lrd->log.redopage.type |= cpu_to_le16(LOG_BTROOT);
		p = &JFS_IP(ip)->i_xtroot;
		if (S_ISDIR(ip->i_mode))
			lrd->log.redopage.type |=
			    cpu_to_le16(LOG_DIR_XTREE);
	} else
		p = (xtpage_t *) mp->data;
	next = le16_to_cpu(p->header.nextindex);

	xtlck = (struct xtlock *) & tlck->lock;

	maplock = (struct maplock *) & tlck->lock;
	xadlock = (struct xdlistlock *) maplock;

	/*
	 *      entry insertion/extension;
	 *      sibling page link update (old right page before split);
	 */
	if (tlck->type & (tlckNEW | tlckGROW | tlckRELINK)) {
		/* log after-image for logredo():
		 * logredo() will update bmap for alloc of new/extended
		 * extents (XAD_NEW|XAD_EXTEND) of XAD[lwm:next) from
		 * after-image of XADlist;
		 * logredo() resets (XAD_NEW|XAD_EXTEND) flag when
		 * applying the after-image to the meta-data page.
		 */
		lrd->type = cpu_to_le16(LOG_REDOPAGE);
1734 1735
		PXDaddress(page_pxd, mp->index);
		PXDlength(page_pxd,
L
Linus Torvalds 已提交
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756
			  mp->logical_size >> tblk->sb->s_blocksize_bits);
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/* format a maplock for txUpdateMap() to update bPMAP
		 * for alloc of new/extended extents of XAD[lwm:next)
		 * from the page itself;
		 * txUpdateMap() resets (XAD_NEW|XAD_EXTEND) flag.
		 */
		lwm = xtlck->lwm.offset;
		if (lwm == 0)
			lwm = XTPAGEMAXSLOT;

		if (lwm == next)
			goto out;
		if (lwm > next) {
			jfs_err("xtLog: lwm > next\n");
			goto out;
		}
		tlck->flag |= tlckUPDATEMAP;
		xadlock->flag = mlckALLOCXADLIST;
		xadlock->count = next - lwm;
1757
		if ((xadlock->count <= 4) && (tblk->xflag & COMMIT_LAZY)) {
L
Linus Torvalds 已提交
1758
			int i;
1759
			pxd_t *pxd;
L
Linus Torvalds 已提交
1760 1761 1762 1763
			/*
			 * Lazy commit may allow xtree to be modified before
			 * txUpdateMap runs.  Copy xad into linelock to
			 * preserve correct data.
1764 1765
			 *
			 * We can fit twice as may pxd's as xads in the lock
L
Linus Torvalds 已提交
1766
			 */
1767 1768 1769 1770 1771
			xadlock->flag = mlckALLOCPXDLIST;
			pxd = xadlock->xdlist = &xtlck->pxdlock;
			for (i = 0; i < xadlock->count; i++) {
				PXDaddress(pxd, addressXAD(&p->xad[lwm + i]));
				PXDlength(pxd, lengthXAD(&p->xad[lwm + i]));
L
Linus Torvalds 已提交
1772 1773
				p->xad[lwm + i].flag &=
				    ~(XAD_NEW | XAD_EXTENDED);
1774 1775
				pxd++;
			}
L
Linus Torvalds 已提交
1776 1777 1778 1779 1780
		} else {
			/*
			 * xdlist will point to into inode's xtree, ensure
			 * that transaction is not committed lazily.
			 */
1781
			xadlock->flag = mlckALLOCXADLIST;
L
Linus Torvalds 已提交
1782 1783 1784 1785 1786 1787 1788 1789 1790 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
			xadlock->xdlist = &p->xad[lwm];
			tblk->xflag &= ~COMMIT_LAZY;
		}
		jfs_info("xtLog: alloc ip:0x%p mp:0x%p tlck:0x%p lwm:%d "
			 "count:%d", tlck->ip, mp, tlck, lwm, xadlock->count);

		maplock->index = 1;

	      out:
		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;

		return;
	}

	/*
	 *      page deletion: file deletion/truncation (ref. xtTruncate())
	 *
	 * (page will be invalidated after log is written and bmap
	 * is updated from the page);
	 */
	if (tlck->type & tlckFREE) {
		/* LOG_NOREDOPAGE log for NoRedoPage filter:
		 * if page free from file delete, NoRedoFile filter from
		 * inode image of zero link count will subsume NoRedoPage
		 * filters for each page;
		 * if page free from file truncattion, write NoRedoPage
		 * filter;
		 *
		 * upadte of block allocation map for the page itself:
		 * if page free from deletion and truncation, LOG_UPDATEMAP
		 * log for the page itself is generated from processing
		 * its parent page xad entries;
		 */
		/* if page free from file truncation, log LOG_NOREDOPAGE
		 * of the deleted page for logredo() to start NoRedoPage
		 * filter for the page;
		 */
		if (tblk->xflag & COMMIT_TRUNCATE) {
			/* write NOREDOPAGE for the page */
			lrd->type = cpu_to_le16(LOG_NOREDOPAGE);
1823 1824
			PXDaddress(page_pxd, mp->index);
			PXDlength(page_pxd,
L
Linus Torvalds 已提交
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
				  mp->logical_size >> tblk->sb->
				  s_blocksize_bits);
			lrd->backchain =
			    cpu_to_le32(lmLog(log, tblk, lrd, NULL));

			if (tlck->type & tlckBTROOT) {
				/* Empty xtree must be logged */
				lrd->type = cpu_to_le16(LOG_REDOPAGE);
				lrd->backchain =
				    cpu_to_le32(lmLog(log, tblk, lrd, tlck));
			}
		}

		/* init LOG_UPDATEMAP of the freed extents
		 * XAD[XTENTRYSTART:hwm) from the deleted page itself
		 * for logredo() to update bmap;
		 */
		lrd->type = cpu_to_le16(LOG_UPDATEMAP);
		lrd->log.updatemap.type = cpu_to_le16(LOG_FREEXADLIST);
		xtlck = (struct xtlock *) & tlck->lock;
		hwm = xtlck->hwm.offset;
		lrd->log.updatemap.nxd =
		    cpu_to_le16(hwm - XTENTRYSTART + 1);
		/* reformat linelock for lmLog() */
		xtlck->header.offset = XTENTRYSTART;
		xtlck->header.length = hwm - XTENTRYSTART + 1;
		xtlck->index = 1;
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/* format a maplock for txUpdateMap() to update bmap
		 * to free extents of XAD[XTENTRYSTART:hwm) from the
		 * deleted page itself;
		 */
		tlck->flag |= tlckUPDATEMAP;
		xadlock->count = hwm - XTENTRYSTART + 1;
1860 1861 1862
		if ((xadlock->count <= 4) && (tblk->xflag & COMMIT_LAZY)) {
			int i;
			pxd_t *pxd;
L
Linus Torvalds 已提交
1863 1864 1865 1866
			/*
			 * Lazy commit may allow xtree to be modified before
			 * txUpdateMap runs.  Copy xad into linelock to
			 * preserve correct data.
1867 1868
			 *
			 * We can fit twice as may pxd's as xads in the lock
L
Linus Torvalds 已提交
1869
			 */
1870 1871 1872 1873 1874 1875 1876 1877 1878
			xadlock->flag = mlckFREEPXDLIST;
			pxd = xadlock->xdlist = &xtlck->pxdlock;
			for (i = 0; i < xadlock->count; i++) {
				PXDaddress(pxd,
					addressXAD(&p->xad[XTENTRYSTART + i]));
				PXDlength(pxd,
					lengthXAD(&p->xad[XTENTRYSTART + i]));
				pxd++;
			}
L
Linus Torvalds 已提交
1879 1880 1881 1882 1883
		} else {
			/*
			 * xdlist will point to into inode's xtree, ensure
			 * that transaction is not committed lazily.
			 */
1884
			xadlock->flag = mlckFREEXADLIST;
L
Linus Torvalds 已提交
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
			xadlock->xdlist = &p->xad[XTENTRYSTART];
			tblk->xflag &= ~COMMIT_LAZY;
		}
		jfs_info("xtLog: free ip:0x%p mp:0x%p count:%d lwm:2",
			 tlck->ip, mp, xadlock->count);

		maplock->index = 1;

		/* mark page as invalid */
		if (((tblk->xflag & COMMIT_PWMAP) || S_ISDIR(ip->i_mode))
		    && !(tlck->type & tlckBTROOT))
			tlck->flag |= tlckFREEPAGE;
		/*
		   else (tblk->xflag & COMMIT_PMAP)
		   ? release the page;
		 */
		return;
	}

	/*
	 *      page/entry truncation: file truncation (ref. xtTruncate())
	 *
	 *     |----------+------+------+---------------|
	 *                |      |      |
	 *                |      |     hwm - hwm before truncation
	 *                |     next - truncation point
	 *               lwm - lwm before truncation
	 * header ?
	 */
	if (tlck->type & tlckTRUNCATE) {
1915
		pxd_t pxd;	/* truncated extent of xad */
L
Linus Torvalds 已提交
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
		int twm;

		/*
		 * For truncation the entire linelock may be used, so it would
		 * be difficult to store xad list in linelock itself.
		 * Therefore, we'll just force transaction to be committed
		 * synchronously, so that xtree pages won't be changed before
		 * txUpdateMap runs.
		 */
		tblk->xflag &= ~COMMIT_LAZY;
		lwm = xtlck->lwm.offset;
		if (lwm == 0)
			lwm = XTPAGEMAXSLOT;
		hwm = xtlck->hwm.offset;
		twm = xtlck->twm.offset;

		/*
		 *      write log records
		 */
		/* log after-image for logredo():
		 *
		 * logredo() will update bmap for alloc of new/extended
		 * extents (XAD_NEW|XAD_EXTEND) of XAD[lwm:next) from
		 * after-image of XADlist;
		 * logredo() resets (XAD_NEW|XAD_EXTEND) flag when
		 * applying the after-image to the meta-data page.
		 */
		lrd->type = cpu_to_le16(LOG_REDOPAGE);
1944 1945 1946
		PXDaddress(page_pxd, mp->index);
		PXDlength(page_pxd,
			  mp->logical_size >> tblk->sb->s_blocksize_bits);
L
Linus Torvalds 已提交
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, tlck));

		/*
		 * truncate entry XAD[twm == next - 1]:
		 */
		if (twm == next - 1) {
			/* init LOG_UPDATEMAP for logredo() to update bmap for
			 * free of truncated delta extent of the truncated
			 * entry XAD[next - 1]:
			 * (xtlck->pxdlock = truncated delta extent);
			 */
			pxdlock = (struct pxd_lock *) & xtlck->pxdlock;
			/* assert(pxdlock->type & tlckTRUNCATE); */
			lrd->type = cpu_to_le16(LOG_UPDATEMAP);
			lrd->log.updatemap.type = cpu_to_le16(LOG_FREEPXD);
			lrd->log.updatemap.nxd = cpu_to_le16(1);
			lrd->log.updatemap.pxd = pxdlock->pxd;
1964
			pxd = pxdlock->pxd;	/* save to format maplock */
L
Linus Torvalds 已提交
1965 1966 1967 1968 1969 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
			lrd->backchain =
			    cpu_to_le32(lmLog(log, tblk, lrd, NULL));
		}

		/*
		 * free entries XAD[next:hwm]:
		 */
		if (hwm >= next) {
			/* init LOG_UPDATEMAP of the freed extents
			 * XAD[next:hwm] from the deleted page itself
			 * for logredo() to update bmap;
			 */
			lrd->type = cpu_to_le16(LOG_UPDATEMAP);
			lrd->log.updatemap.type =
			    cpu_to_le16(LOG_FREEXADLIST);
			xtlck = (struct xtlock *) & tlck->lock;
			hwm = xtlck->hwm.offset;
			lrd->log.updatemap.nxd =
			    cpu_to_le16(hwm - next + 1);
			/* reformat linelock for lmLog() */
			xtlck->header.offset = next;
			xtlck->header.length = hwm - next + 1;
			xtlck->index = 1;
			lrd->backchain =
			    cpu_to_le32(lmLog(log, tblk, lrd, tlck));
		}

		/*
		 *      format maplock(s) for txUpdateMap() to update bmap
		 */
		maplock->index = 0;

		/*
		 * allocate entries XAD[lwm:next):
		 */
		if (lwm < next) {
			/* format a maplock for txUpdateMap() to update bPMAP
			 * for alloc of new/extended extents of XAD[lwm:next)
			 * from the page itself;
			 * txUpdateMap() resets (XAD_NEW|XAD_EXTEND) flag.
			 */
			tlck->flag |= tlckUPDATEMAP;
			xadlock->flag = mlckALLOCXADLIST;
			xadlock->count = next - lwm;
			xadlock->xdlist = &p->xad[lwm];

			jfs_info("xtLog: alloc ip:0x%p mp:0x%p count:%d "
				 "lwm:%d next:%d",
				 tlck->ip, mp, xadlock->count, lwm, next);
			maplock->index++;
			xadlock++;
		}

		/*
		 * truncate entry XAD[twm == next - 1]:
		 */
		if (twm == next - 1) {
			struct pxd_lock *pxdlock;

			/* format a maplock for txUpdateMap() to update bmap
			 * to free truncated delta extent of the truncated
			 * entry XAD[next - 1];
			 * (xtlck->pxdlock = truncated delta extent);
			 */
			tlck->flag |= tlckUPDATEMAP;
			pxdlock = (struct pxd_lock *) xadlock;
			pxdlock->flag = mlckFREEPXD;
			pxdlock->count = 1;
2033
			pxdlock->pxd = pxd;
L
Linus Torvalds 已提交
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 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 2180 2181 2182 2183 2184 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

			jfs_info("xtLog: truncate ip:0x%p mp:0x%p count:%d "
				 "hwm:%d", ip, mp, pxdlock->count, hwm);
			maplock->index++;
			xadlock++;
		}

		/*
		 * free entries XAD[next:hwm]:
		 */
		if (hwm >= next) {
			/* format a maplock for txUpdateMap() to update bmap
			 * to free extents of XAD[next:hwm] from thedeleted
			 * page itself;
			 */
			tlck->flag |= tlckUPDATEMAP;
			xadlock->flag = mlckFREEXADLIST;
			xadlock->count = hwm - next + 1;
			xadlock->xdlist = &p->xad[next];

			jfs_info("xtLog: free ip:0x%p mp:0x%p count:%d "
				 "next:%d hwm:%d",
				 tlck->ip, mp, xadlock->count, next, hwm);
			maplock->index++;
		}

		/* mark page as homeward bound */
		tlck->flag |= tlckWRITEPAGE;
	}
	return;
}

/*
 *      mapLog()
 *
 * function:    log from maplock of freed data extents;
 */
void mapLog(struct jfs_log * log, struct tblock * tblk, struct lrd * lrd,
	    struct tlock * tlck)
{
	struct pxd_lock *pxdlock;
	int i, nlock;
	pxd_t *pxd;

	/*
	 *      page relocation: free the source page extent
	 *
	 * a maplock for txUpdateMap() for free of the page
	 * has been formatted at txLock() time saving the src
	 * relocated page address;
	 */
	if (tlck->type & tlckRELOCATE) {
		/* log LOG_NOREDOPAGE of the old relocated page
		 * for logredo() to start NoRedoPage filter;
		 */
		lrd->type = cpu_to_le16(LOG_NOREDOPAGE);
		pxdlock = (struct pxd_lock *) & tlck->lock;
		pxd = &lrd->log.redopage.pxd;
		*pxd = pxdlock->pxd;
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, NULL));

		/* (N.B. currently, logredo() does NOT update bmap
		 * for free of the page itself for (LOG_XTREE|LOG_NOREDOPAGE);
		 * if page free from relocation, LOG_UPDATEMAP log is
		 * specifically generated now for logredo()
		 * to update bmap for free of src relocated page;
		 * (new flag LOG_RELOCATE may be introduced which will
		 * inform logredo() to start NORedoPage filter and also
		 * update block allocation map at the same time, thus
		 * avoiding an extra log write);
		 */
		lrd->type = cpu_to_le16(LOG_UPDATEMAP);
		lrd->log.updatemap.type = cpu_to_le16(LOG_FREEPXD);
		lrd->log.updatemap.nxd = cpu_to_le16(1);
		lrd->log.updatemap.pxd = pxdlock->pxd;
		lrd->backchain = cpu_to_le32(lmLog(log, tblk, lrd, NULL));

		/* a maplock for txUpdateMap() for free of the page
		 * has been formatted at txLock() time;
		 */
		tlck->flag |= tlckUPDATEMAP;
		return;
	}
	/*

	 * Otherwise it's not a relocate request
	 *
	 */
	else {
		/* log LOG_UPDATEMAP for logredo() to update bmap for
		 * free of truncated/relocated delta extent of the data;
		 * e.g.: external EA extent, relocated/truncated extent
		 * from xtTailgate();
		 */
		lrd->type = cpu_to_le16(LOG_UPDATEMAP);
		pxdlock = (struct pxd_lock *) & tlck->lock;
		nlock = pxdlock->index;
		for (i = 0; i < nlock; i++, pxdlock++) {
			if (pxdlock->flag & mlckALLOCPXD)
				lrd->log.updatemap.type =
				    cpu_to_le16(LOG_ALLOCPXD);
			else
				lrd->log.updatemap.type =
				    cpu_to_le16(LOG_FREEPXD);
			lrd->log.updatemap.nxd = cpu_to_le16(1);
			lrd->log.updatemap.pxd = pxdlock->pxd;
			lrd->backchain =
			    cpu_to_le32(lmLog(log, tblk, lrd, NULL));
			jfs_info("mapLog: xaddr:0x%lx xlen:0x%x",
				 (ulong) addressPXD(&pxdlock->pxd),
				 lengthPXD(&pxdlock->pxd));
		}

		/* update bmap */
		tlck->flag |= tlckUPDATEMAP;
	}
}

/*
 *      txEA()
 *
 * function:    acquire maplock for EA/ACL extents or
 *              set COMMIT_INLINE flag;
 */
void txEA(tid_t tid, struct inode *ip, dxd_t * oldea, dxd_t * newea)
{
	struct tlock *tlck = NULL;
	struct pxd_lock *maplock = NULL, *pxdlock = NULL;

	/*
	 * format maplock for alloc of new EA extent
	 */
	if (newea) {
		/* Since the newea could be a completely zeroed entry we need to
		 * check for the two flags which indicate we should actually
		 * commit new EA data
		 */
		if (newea->flag & DXD_EXTENT) {
			tlck = txMaplock(tid, ip, tlckMAP);
			maplock = (struct pxd_lock *) & tlck->lock;
			pxdlock = (struct pxd_lock *) maplock;
			pxdlock->flag = mlckALLOCPXD;
			PXDaddress(&pxdlock->pxd, addressDXD(newea));
			PXDlength(&pxdlock->pxd, lengthDXD(newea));
			pxdlock++;
			maplock->index = 1;
		} else if (newea->flag & DXD_INLINE) {
			tlck = NULL;

			set_cflag(COMMIT_Inlineea, ip);
		}
	}

	/*
	 * format maplock for free of old EA extent
	 */
	if (!test_cflag(COMMIT_Nolink, ip) && oldea->flag & DXD_EXTENT) {
		if (tlck == NULL) {
			tlck = txMaplock(tid, ip, tlckMAP);
			maplock = (struct pxd_lock *) & tlck->lock;
			pxdlock = (struct pxd_lock *) maplock;
			maplock->index = 0;
		}
		pxdlock->flag = mlckFREEPXD;
		PXDaddress(&pxdlock->pxd, addressDXD(oldea));
		PXDlength(&pxdlock->pxd, lengthDXD(oldea));
		maplock->index++;
	}
}

/*
 *      txForce()
 *
 * function: synchronously write pages locked by transaction
 *              after txLog() but before txUpdateMap();
 */
void txForce(struct tblock * tblk)
{
	struct tlock *tlck;
	lid_t lid, next;
	struct metapage *mp;

	/*
	 * reverse the order of transaction tlocks in
	 * careful update order of address index pages
	 * (right to left, bottom up)
	 */
	tlck = lid_to_tlock(tblk->next);
	lid = tlck->next;
	tlck->next = 0;
	while (lid) {
		tlck = lid_to_tlock(lid);
		next = tlck->next;
		tlck->next = tblk->next;
		tblk->next = lid;
		lid = next;
	}

	/*
	 * synchronously write the page, and
	 * hold the page for txUpdateMap();
	 */
	for (lid = tblk->next; lid; lid = next) {
		tlck = lid_to_tlock(lid);
		next = tlck->next;

		if ((mp = tlck->mp) != NULL &&
		    (tlck->type & tlckBTROOT) == 0) {
			assert(mp->xflag & COMMIT_PAGE);

			if (tlck->flag & tlckWRITEPAGE) {
				tlck->flag &= ~tlckWRITEPAGE;

				/* do not release page to freelist */
2248 2249
				force_metapage(mp);
#if 0
L
Linus Torvalds 已提交
2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260
				/*
				 * The "right" thing to do here is to
				 * synchronously write the metadata.
				 * With the current implementation this
				 * is hard since write_metapage requires
				 * us to kunmap & remap the page.  If we
				 * have tlocks pointing into the metadata
				 * pages, we don't want to do this.  I think
				 * we can get by with synchronously writing
				 * the pages when they are released.
				 */
2261
				assert(mp->nohomeok);
L
Linus Torvalds 已提交
2262 2263
				set_bit(META_dirty, &mp->flag);
				set_bit(META_sync, &mp->flag);
2264
#endif
L
Linus Torvalds 已提交
2265 2266 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 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
			}
		}
	}
}

/*
 *      txUpdateMap()
 *
 * function:    update persistent allocation map (and working map
 *              if appropriate);
 *
 * parameter:
 */
static void txUpdateMap(struct tblock * tblk)
{
	struct inode *ip;
	struct inode *ipimap;
	lid_t lid;
	struct tlock *tlck;
	struct maplock *maplock;
	struct pxd_lock pxdlock;
	int maptype;
	int k, nlock;
	struct metapage *mp = NULL;

	ipimap = JFS_SBI(tblk->sb)->ipimap;

	maptype = (tblk->xflag & COMMIT_PMAP) ? COMMIT_PMAP : COMMIT_PWMAP;


	/*
	 *      update block allocation map
	 *
	 * update allocation state in pmap (and wmap) and
	 * update lsn of the pmap page;
	 */
	/*
	 * scan each tlock/page of transaction for block allocation/free:
	 *
	 * for each tlock/page of transaction, update map.
	 *  ? are there tlock for pmap and pwmap at the same time ?
	 */
	for (lid = tblk->next; lid; lid = tlck->next) {
		tlck = lid_to_tlock(lid);

		if ((tlck->flag & tlckUPDATEMAP) == 0)
			continue;

		if (tlck->flag & tlckFREEPAGE) {
			/*
			 * Another thread may attempt to reuse freed space
			 * immediately, so we want to get rid of the metapage
			 * before anyone else has a chance to get it.
			 * Lock metapage, update maps, then invalidate
			 * the metapage.
			 */
			mp = tlck->mp;
			ASSERT(mp->xflag & COMMIT_PAGE);
2323
			grab_metapage(mp);
L
Linus Torvalds 已提交
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
		}

		/*
		 * extent list:
		 * . in-line PXD list:
		 * . out-of-line XAD list:
		 */
		maplock = (struct maplock *) & tlck->lock;
		nlock = maplock->index;

		for (k = 0; k < nlock; k++, maplock++) {
			/*
			 * allocate blocks in persistent map:
			 *
			 * blocks have been allocated from wmap at alloc time;
			 */
			if (maplock->flag & mlckALLOC) {
				txAllocPMap(ipimap, maplock, tblk);
			}
			/*
			 * free blocks in persistent and working map:
			 * blocks will be freed in pmap and then in wmap;
			 *
			 * ? tblock specifies the PMAP/PWMAP based upon
			 * transaction
			 *
			 * free blocks in persistent map:
			 * blocks will be freed from wmap at last reference
			 * release of the object for regular files;
			 *
			 * Alway free blocks from both persistent & working
			 * maps for directories
			 */
			else {	/* (maplock->flag & mlckFREE) */

				if (S_ISDIR(tlck->ip->i_mode))
					txFreeMap(ipimap, maplock,
						  tblk, COMMIT_PWMAP);
				else
					txFreeMap(ipimap, maplock,
						  tblk, maptype);
			}
		}
		if (tlck->flag & tlckFREEPAGE) {
			if (!(tblk->flag & tblkGC_LAZY)) {
				/* This is equivalent to txRelease */
				ASSERT(mp->lid == lid);
				tlck->mp->lid = 0;
			}
2373 2374
			assert(mp->nohomeok == 1);
			metapage_homeok(mp);
L
Linus Torvalds 已提交
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 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 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 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 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 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 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 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 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 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 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 2785 2786 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 2835
			discard_metapage(mp);
			tlck->mp = NULL;
		}
	}
	/*
	 *      update inode allocation map
	 *
	 * update allocation state in pmap and
	 * update lsn of the pmap page;
	 * update in-memory inode flag/state
	 *
	 * unlock mapper/write lock
	 */
	if (tblk->xflag & COMMIT_CREATE) {
		diUpdatePMap(ipimap, tblk->ino, FALSE, tblk);
		ipimap->i_state |= I_DIRTY;
		/* update persistent block allocation map
		 * for the allocation of inode extent;
		 */
		pxdlock.flag = mlckALLOCPXD;
		pxdlock.pxd = tblk->u.ixpxd;
		pxdlock.index = 1;
		txAllocPMap(ipimap, (struct maplock *) & pxdlock, tblk);
	} else if (tblk->xflag & COMMIT_DELETE) {
		ip = tblk->u.ip;
		diUpdatePMap(ipimap, ip->i_ino, TRUE, tblk);
		ipimap->i_state |= I_DIRTY;
		iput(ip);
	}
}

/*
 *      txAllocPMap()
 *
 * function: allocate from persistent map;
 *
 * parameter:
 *      ipbmap  -
 *      malock -
 *              xad list:
 *              pxd:
 *
 *      maptype -
 *              allocate from persistent map;
 *              free from persistent map;
 *              (e.g., tmp file - free from working map at releae
 *               of last reference);
 *              free from persistent and working map;
 *
 *      lsn     - log sequence number;
 */
static void txAllocPMap(struct inode *ip, struct maplock * maplock,
			struct tblock * tblk)
{
	struct inode *ipbmap = JFS_SBI(ip->i_sb)->ipbmap;
	struct xdlistlock *xadlistlock;
	xad_t *xad;
	s64 xaddr;
	int xlen;
	struct pxd_lock *pxdlock;
	struct xdlistlock *pxdlistlock;
	pxd_t *pxd;
	int n;

	/*
	 * allocate from persistent map;
	 */
	if (maplock->flag & mlckALLOCXADLIST) {
		xadlistlock = (struct xdlistlock *) maplock;
		xad = xadlistlock->xdlist;
		for (n = 0; n < xadlistlock->count; n++, xad++) {
			if (xad->flag & (XAD_NEW | XAD_EXTENDED)) {
				xaddr = addressXAD(xad);
				xlen = lengthXAD(xad);
				dbUpdatePMap(ipbmap, FALSE, xaddr,
					     (s64) xlen, tblk);
				xad->flag &= ~(XAD_NEW | XAD_EXTENDED);
				jfs_info("allocPMap: xaddr:0x%lx xlen:%d",
					 (ulong) xaddr, xlen);
			}
		}
	} else if (maplock->flag & mlckALLOCPXD) {
		pxdlock = (struct pxd_lock *) maplock;
		xaddr = addressPXD(&pxdlock->pxd);
		xlen = lengthPXD(&pxdlock->pxd);
		dbUpdatePMap(ipbmap, FALSE, xaddr, (s64) xlen, tblk);
		jfs_info("allocPMap: xaddr:0x%lx xlen:%d", (ulong) xaddr, xlen);
	} else {		/* (maplock->flag & mlckALLOCPXDLIST) */

		pxdlistlock = (struct xdlistlock *) maplock;
		pxd = pxdlistlock->xdlist;
		for (n = 0; n < pxdlistlock->count; n++, pxd++) {
			xaddr = addressPXD(pxd);
			xlen = lengthPXD(pxd);
			dbUpdatePMap(ipbmap, FALSE, xaddr, (s64) xlen,
				     tblk);
			jfs_info("allocPMap: xaddr:0x%lx xlen:%d",
				 (ulong) xaddr, xlen);
		}
	}
}

/*
 *      txFreeMap()
 *
 * function:    free from persistent and/or working map;
 *
 * todo: optimization
 */
void txFreeMap(struct inode *ip,
	       struct maplock * maplock, struct tblock * tblk, int maptype)
{
	struct inode *ipbmap = JFS_SBI(ip->i_sb)->ipbmap;
	struct xdlistlock *xadlistlock;
	xad_t *xad;
	s64 xaddr;
	int xlen;
	struct pxd_lock *pxdlock;
	struct xdlistlock *pxdlistlock;
	pxd_t *pxd;
	int n;

	jfs_info("txFreeMap: tblk:0x%p maplock:0x%p maptype:0x%x",
		 tblk, maplock, maptype);

	/*
	 * free from persistent map;
	 */
	if (maptype == COMMIT_PMAP || maptype == COMMIT_PWMAP) {
		if (maplock->flag & mlckFREEXADLIST) {
			xadlistlock = (struct xdlistlock *) maplock;
			xad = xadlistlock->xdlist;
			for (n = 0; n < xadlistlock->count; n++, xad++) {
				if (!(xad->flag & XAD_NEW)) {
					xaddr = addressXAD(xad);
					xlen = lengthXAD(xad);
					dbUpdatePMap(ipbmap, TRUE, xaddr,
						     (s64) xlen, tblk);
					jfs_info("freePMap: xaddr:0x%lx "
						 "xlen:%d",
						 (ulong) xaddr, xlen);
				}
			}
		} else if (maplock->flag & mlckFREEPXD) {
			pxdlock = (struct pxd_lock *) maplock;
			xaddr = addressPXD(&pxdlock->pxd);
			xlen = lengthPXD(&pxdlock->pxd);
			dbUpdatePMap(ipbmap, TRUE, xaddr, (s64) xlen,
				     tblk);
			jfs_info("freePMap: xaddr:0x%lx xlen:%d",
				 (ulong) xaddr, xlen);
		} else {	/* (maplock->flag & mlckALLOCPXDLIST) */

			pxdlistlock = (struct xdlistlock *) maplock;
			pxd = pxdlistlock->xdlist;
			for (n = 0; n < pxdlistlock->count; n++, pxd++) {
				xaddr = addressPXD(pxd);
				xlen = lengthPXD(pxd);
				dbUpdatePMap(ipbmap, TRUE, xaddr,
					     (s64) xlen, tblk);
				jfs_info("freePMap: xaddr:0x%lx xlen:%d",
					 (ulong) xaddr, xlen);
			}
		}
	}

	/*
	 * free from working map;
	 */
	if (maptype == COMMIT_PWMAP || maptype == COMMIT_WMAP) {
		if (maplock->flag & mlckFREEXADLIST) {
			xadlistlock = (struct xdlistlock *) maplock;
			xad = xadlistlock->xdlist;
			for (n = 0; n < xadlistlock->count; n++, xad++) {
				xaddr = addressXAD(xad);
				xlen = lengthXAD(xad);
				dbFree(ip, xaddr, (s64) xlen);
				xad->flag = 0;
				jfs_info("freeWMap: xaddr:0x%lx xlen:%d",
					 (ulong) xaddr, xlen);
			}
		} else if (maplock->flag & mlckFREEPXD) {
			pxdlock = (struct pxd_lock *) maplock;
			xaddr = addressPXD(&pxdlock->pxd);
			xlen = lengthPXD(&pxdlock->pxd);
			dbFree(ip, xaddr, (s64) xlen);
			jfs_info("freeWMap: xaddr:0x%lx xlen:%d",
				 (ulong) xaddr, xlen);
		} else {	/* (maplock->flag & mlckFREEPXDLIST) */

			pxdlistlock = (struct xdlistlock *) maplock;
			pxd = pxdlistlock->xdlist;
			for (n = 0; n < pxdlistlock->count; n++, pxd++) {
				xaddr = addressPXD(pxd);
				xlen = lengthPXD(pxd);
				dbFree(ip, xaddr, (s64) xlen);
				jfs_info("freeWMap: xaddr:0x%lx xlen:%d",
					 (ulong) xaddr, xlen);
			}
		}
	}
}

/*
 *      txFreelock()
 *
 * function:    remove tlock from inode anonymous locklist
 */
void txFreelock(struct inode *ip)
{
	struct jfs_inode_info *jfs_ip = JFS_IP(ip);
	struct tlock *xtlck, *tlck;
	lid_t xlid = 0, lid;

	if (!jfs_ip->atlhead)
		return;

	TXN_LOCK();
	xtlck = (struct tlock *) &jfs_ip->atlhead;

	while ((lid = xtlck->next) != 0) {
		tlck = lid_to_tlock(lid);
		if (tlck->flag & tlckFREELOCK) {
			xtlck->next = tlck->next;
			txLockFree(lid);
		} else {
			xtlck = tlck;
			xlid = lid;
		}
	}

	if (jfs_ip->atlhead)
		jfs_ip->atltail = xlid;
	else {
		jfs_ip->atltail = 0;
		/*
		 * If inode was on anon_list, remove it
		 */
		list_del_init(&jfs_ip->anon_inode_list);
	}
	TXN_UNLOCK();
}

/*
 *      txAbort()
 *
 * function: abort tx before commit;
 *
 * frees line-locks and segment locks for all
 * segments in comdata structure.
 * Optionally sets state of file-system to FM_DIRTY in super-block.
 * log age of page-frames in memory for which caller has
 * are reset to 0 (to avoid logwarap).
 */
void txAbort(tid_t tid, int dirty)
{
	lid_t lid, next;
	struct metapage *mp;
	struct tblock *tblk = tid_to_tblock(tid);
	struct tlock *tlck;

	/*
	 * free tlocks of the transaction
	 */
	for (lid = tblk->next; lid; lid = next) {
		tlck = lid_to_tlock(lid);
		next = tlck->next;
		mp = tlck->mp;
		JFS_IP(tlck->ip)->xtlid = 0;

		if (mp) {
			mp->lid = 0;

			/*
			 * reset lsn of page to avoid logwarap:
			 *
			 * (page may have been previously committed by another
			 * transaction(s) but has not been paged, i.e.,
			 * it may be on logsync list even though it has not
			 * been logged for the current tx.)
			 */
			if (mp->xflag & COMMIT_PAGE && mp->lsn)
				LogSyncRelease(mp);
		}
		/* insert tlock at head of freelist */
		TXN_LOCK();
		txLockFree(lid);
		TXN_UNLOCK();
	}

	/* caller will free the transaction block */

	tblk->next = tblk->last = 0;

	/*
	 * mark filesystem dirty
	 */
	if (dirty)
		jfs_error(tblk->sb, "txAbort");

	return;
}

/*
 *      txLazyCommit(void)
 *
 *	All transactions except those changing ipimap (COMMIT_FORCE) are
 *	processed by this routine.  This insures that the inode and block
 *	allocation maps are updated in order.  For synchronous transactions,
 *	let the user thread finish processing after txUpdateMap() is called.
 */
static void txLazyCommit(struct tblock * tblk)
{
	struct jfs_log *log;

	while (((tblk->flag & tblkGC_READY) == 0) &&
	       ((tblk->flag & tblkGC_UNLOCKED) == 0)) {
		/* We must have gotten ahead of the user thread
		 */
		jfs_info("jfs_lazycommit: tblk 0x%p not unlocked", tblk);
		yield();
	}

	jfs_info("txLazyCommit: processing tblk 0x%p", tblk);

	txUpdateMap(tblk);

	log = (struct jfs_log *) JFS_SBI(tblk->sb)->log;

	spin_lock_irq(&log->gclock);	// LOGGC_LOCK

	tblk->flag |= tblkGC_COMMITTED;

	if (tblk->flag & tblkGC_READY)
		log->gcrtc--;

	wake_up_all(&tblk->gcwait);	// LOGGC_WAKEUP

	/*
	 * Can't release log->gclock until we've tested tblk->flag
	 */
	if (tblk->flag & tblkGC_LAZY) {
		spin_unlock_irq(&log->gclock);	// LOGGC_UNLOCK
		txUnlock(tblk);
		tblk->flag &= ~tblkGC_LAZY;
		txEnd(tblk - TxBlock);	/* Convert back to tid */
	} else
		spin_unlock_irq(&log->gclock);	// LOGGC_UNLOCK

	jfs_info("txLazyCommit: done: tblk = 0x%p", tblk);
}

/*
 *      jfs_lazycommit(void)
 *
 *	To be run as a kernel daemon.  If lbmIODone is called in an interrupt
 *	context, or where blocking is not wanted, this routine will process
 *	committed transactions from the unlock queue.
 */
int jfs_lazycommit(void *arg)
{
	int WorkDone;
	struct tblock *tblk;
	unsigned long flags;
	struct jfs_sb_info *sbi;

	daemonize("jfsCommit");

	complete(&jfsIOwait);

	do {
		LAZY_LOCK(flags);
		jfs_commit_thread_waking = 0;	/* OK to wake another thread */
		while (!list_empty(&TxAnchor.unlock_queue)) {
			WorkDone = 0;
			list_for_each_entry(tblk, &TxAnchor.unlock_queue,
					    cqueue) {

				sbi = JFS_SBI(tblk->sb);
				/*
				 * For each volume, the transactions must be
				 * handled in order.  If another commit thread
				 * is handling a tblk for this superblock,
				 * skip it
				 */
				if (sbi->commit_state & IN_LAZYCOMMIT)
					continue;

				sbi->commit_state |= IN_LAZYCOMMIT;
				WorkDone = 1;

				/*
				 * Remove transaction from queue
				 */
				list_del(&tblk->cqueue);

				LAZY_UNLOCK(flags);
				txLazyCommit(tblk);
				LAZY_LOCK(flags);

				sbi->commit_state &= ~IN_LAZYCOMMIT;
				/*
				 * Don't continue in the for loop.  (We can't
				 * anyway, it's unsafe!)  We want to go back to
				 * the beginning of the list.
				 */
				break;
			}

			/* If there was nothing to do, don't continue */
			if (!WorkDone)
				break;
		}
		/* In case a wakeup came while all threads were active */
		jfs_commit_thread_waking = 0;

		if (current->flags & PF_FREEZE) {
			LAZY_UNLOCK(flags);
			refrigerator(PF_FREEZE);
		} else {
			DECLARE_WAITQUEUE(wq, current);

			add_wait_queue(&jfs_commit_thread_wait, &wq);
			set_current_state(TASK_INTERRUPTIBLE);
			LAZY_UNLOCK(flags);
			schedule();
			current->state = TASK_RUNNING;
			remove_wait_queue(&jfs_commit_thread_wait, &wq);
		}
	} while (!jfs_stop_threads);

	if (!list_empty(&TxAnchor.unlock_queue))
		jfs_err("jfs_lazycommit being killed w/pending transactions!");
	else
		jfs_info("jfs_lazycommit being killed\n");
	complete_and_exit(&jfsIOwait, 0);
}

void txLazyUnlock(struct tblock * tblk)
{
	unsigned long flags;

	LAZY_LOCK(flags);

	list_add_tail(&tblk->cqueue, &TxAnchor.unlock_queue);
	/*
	 * Don't wake up a commit thread if there is already one servicing
	 * this superblock, or if the last one we woke up hasn't started yet.
	 */
	if (!(JFS_SBI(tblk->sb)->commit_state & IN_LAZYCOMMIT) &&
	    !jfs_commit_thread_waking) {
		jfs_commit_thread_waking = 1;
		wake_up(&jfs_commit_thread_wait);
	}
	LAZY_UNLOCK(flags);
}

static void LogSyncRelease(struct metapage * mp)
{
	struct jfs_log *log = mp->log;

2836
	assert(mp->nohomeok);
L
Linus Torvalds 已提交
2837
	assert(log);
2838
	metapage_homeok(mp);
L
Linus Torvalds 已提交
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 2873 2874 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 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 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 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105
}

/*
 *	txQuiesce
 *
 *	Block all new transactions and push anonymous transactions to
 *	completion
 *
 *	This does almost the same thing as jfs_sync below.  We don't
 *	worry about deadlocking when jfs_tlocks_low is set, since we would
 *	expect jfs_sync to get us out of that jam.
 */
void txQuiesce(struct super_block *sb)
{
	struct inode *ip;
	struct jfs_inode_info *jfs_ip;
	struct jfs_log *log = JFS_SBI(sb)->log;
	tid_t tid;

	set_bit(log_QUIESCE, &log->flag);

	TXN_LOCK();
restart:
	while (!list_empty(&TxAnchor.anon_list)) {
		jfs_ip = list_entry(TxAnchor.anon_list.next,
				    struct jfs_inode_info,
				    anon_inode_list);
		ip = &jfs_ip->vfs_inode;

		/*
		 * inode will be removed from anonymous list
		 * when it is committed
		 */
		TXN_UNLOCK();
		tid = txBegin(ip->i_sb, COMMIT_INODE | COMMIT_FORCE);
		down(&jfs_ip->commit_sem);
		txCommit(tid, 1, &ip, 0);
		txEnd(tid);
		up(&jfs_ip->commit_sem);
		/*
		 * Just to be safe.  I don't know how
		 * long we can run without blocking
		 */
		cond_resched();
		TXN_LOCK();
	}

	/*
	 * If jfs_sync is running in parallel, there could be some inodes
	 * on anon_list2.  Let's check.
	 */
	if (!list_empty(&TxAnchor.anon_list2)) {
		list_splice(&TxAnchor.anon_list2, &TxAnchor.anon_list);
		INIT_LIST_HEAD(&TxAnchor.anon_list2);
		goto restart;
	}
	TXN_UNLOCK();

	/*
	 * We may need to kick off the group commit
	 */
	jfs_flush_journal(log, 0);
}

/*
 * txResume()
 *
 * Allows transactions to start again following txQuiesce
 */
void txResume(struct super_block *sb)
{
	struct jfs_log *log = JFS_SBI(sb)->log;

	clear_bit(log_QUIESCE, &log->flag);
	TXN_WAKEUP(&log->syncwait);
}

/*
 *      jfs_sync(void)
 *
 *	To be run as a kernel daemon.  This is awakened when tlocks run low.
 *	We write any inodes that have anonymous tlocks so they will become
 *	available.
 */
int jfs_sync(void *arg)
{
	struct inode *ip;
	struct jfs_inode_info *jfs_ip;
	int rc;
	tid_t tid;

	daemonize("jfsSync");

	complete(&jfsIOwait);

	do {
		/*
		 * write each inode on the anonymous inode list
		 */
		TXN_LOCK();
		while (jfs_tlocks_low && !list_empty(&TxAnchor.anon_list)) {
			jfs_ip = list_entry(TxAnchor.anon_list.next,
					    struct jfs_inode_info,
					    anon_inode_list);
			ip = &jfs_ip->vfs_inode;

			if (! igrab(ip)) {
				/*
				 * Inode is being freed
				 */
				list_del_init(&jfs_ip->anon_inode_list);
			} else if (! down_trylock(&jfs_ip->commit_sem)) {
				/*
				 * inode will be removed from anonymous list
				 * when it is committed
				 */
				TXN_UNLOCK();
				tid = txBegin(ip->i_sb, COMMIT_INODE);
				rc = txCommit(tid, 1, &ip, 0);
				txEnd(tid);
				up(&jfs_ip->commit_sem);

				iput(ip);
				/*
				 * Just to be safe.  I don't know how
				 * long we can run without blocking
				 */
				cond_resched();
				TXN_LOCK();
			} else {
				/* We can't get the commit semaphore.  It may
				 * be held by a thread waiting for tlock's
				 * so let's not block here.  Save it to
				 * put back on the anon_list.
				 */

				/* Take off anon_list */
				list_del(&jfs_ip->anon_inode_list);

				/* Put on anon_list2 */
				list_add(&jfs_ip->anon_inode_list,
					 &TxAnchor.anon_list2);

				TXN_UNLOCK();
				iput(ip);
				TXN_LOCK();
			}
		}
		/* Add anon_list2 back to anon_list */
		list_splice_init(&TxAnchor.anon_list2, &TxAnchor.anon_list);

		if (current->flags & PF_FREEZE) {
			TXN_UNLOCK();
			refrigerator(PF_FREEZE);
		} else {
			DECLARE_WAITQUEUE(wq, current);

			add_wait_queue(&jfs_sync_thread_wait, &wq);
			set_current_state(TASK_INTERRUPTIBLE);
			TXN_UNLOCK();
			schedule();
			current->state = TASK_RUNNING;
			remove_wait_queue(&jfs_sync_thread_wait, &wq);
		}
	} while (!jfs_stop_threads);

	jfs_info("jfs_sync being killed");
	complete_and_exit(&jfsIOwait, 0);
}

#if defined(CONFIG_PROC_FS) && defined(CONFIG_JFS_DEBUG)
int jfs_txanchor_read(char *buffer, char **start, off_t offset, int length,
		      int *eof, void *data)
{
	int len = 0;
	off_t begin;
	char *freewait;
	char *freelockwait;
	char *lowlockwait;

	freewait =
	    waitqueue_active(&TxAnchor.freewait) ? "active" : "empty";
	freelockwait =
	    waitqueue_active(&TxAnchor.freelockwait) ? "active" : "empty";
	lowlockwait =
	    waitqueue_active(&TxAnchor.lowlockwait) ? "active" : "empty";

	len += sprintf(buffer,
		       "JFS TxAnchor\n"
		       "============\n"
		       "freetid = %d\n"
		       "freewait = %s\n"
		       "freelock = %d\n"
		       "freelockwait = %s\n"
		       "lowlockwait = %s\n"
		       "tlocksInUse = %d\n"
		       "jfs_tlocks_low = %d\n"
		       "unlock_queue is %sempty\n",
		       TxAnchor.freetid,
		       freewait,
		       TxAnchor.freelock,
		       freelockwait,
		       lowlockwait,
		       TxAnchor.tlocksInUse,
		       jfs_tlocks_low,
		       list_empty(&TxAnchor.unlock_queue) ? "" : "not ");

	begin = offset;
	*start = buffer + begin;
	len -= begin;

	if (len > length)
		len = length;
	else
		*eof = 1;

	if (len < 0)
		len = 0;

	return len;
}
#endif

#if defined(CONFIG_PROC_FS) && defined(CONFIG_JFS_STATISTICS)
int jfs_txstats_read(char *buffer, char **start, off_t offset, int length,
		     int *eof, void *data)
{
	int len = 0;
	off_t begin;

	len += sprintf(buffer,
		       "JFS TxStats\n"
		       "===========\n"
		       "calls to txBegin = %d\n"
		       "txBegin blocked by sync barrier = %d\n"
		       "txBegin blocked by tlocks low = %d\n"
		       "txBegin blocked by no free tid = %d\n"
		       "calls to txBeginAnon = %d\n"
		       "txBeginAnon blocked by sync barrier = %d\n"
		       "txBeginAnon blocked by tlocks low = %d\n"
		       "calls to txLockAlloc = %d\n"
		       "tLockAlloc blocked by no free lock = %d\n",
		       TxStat.txBegin,
		       TxStat.txBegin_barrier,
		       TxStat.txBegin_lockslow,
		       TxStat.txBegin_freetid,
		       TxStat.txBeginAnon,
		       TxStat.txBeginAnon_barrier,
		       TxStat.txBeginAnon_lockslow,
		       TxStat.txLockAlloc,
		       TxStat.txLockAlloc_freelock);

	begin = offset;
	*start = buffer + begin;
	len -= begin;

	if (len > length)
		len = length;
	else
		*eof = 1;

	if (len < 0)
		len = 0;

	return len;
}
#endif