xfs_sync.c 18.6 KB
Newer Older
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
/*
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * 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.
 *
 * This program is distributed in the hope that it would 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 the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_types.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_dir2.h"
#include "xfs_dmapi.h"
#include "xfs_mount.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_btree.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_inode.h"
#include "xfs_dinode.h"
#include "xfs_error.h"
#include "xfs_mru_cache.h"
#include "xfs_filestream.h"
#include "xfs_vnodeops.h"
#include "xfs_utils.h"
#include "xfs_buf_item.h"
#include "xfs_inode_item.h"
#include "xfs_rw.h"
C
Christoph Hellwig 已提交
46
#include "xfs_quota.h"
47

48 49 50
#include <linux/kthread.h>
#include <linux/freezer.h>

51

52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
/* must be called with pag_ici_lock held and releases it */
STATIC int
xfs_sync_inode_valid(
	struct xfs_inode	*ip,
	struct xfs_perag	*pag)
{
	struct inode		*inode = VFS_I(ip);

	/* nothing to sync during shutdown */
	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
		read_unlock(&pag->pag_ici_lock);
		return EFSCORRUPTED;
	}

	/*
	 * If we can't get a reference on the inode, it must be in reclaim.
	 * Leave it for the reclaim code to flush. Also avoid inodes that
	 * haven't been fully initialised.
	 */
	if (!igrab(inode)) {
		read_unlock(&pag->pag_ici_lock);
		return ENOENT;
	}
	read_unlock(&pag->pag_ici_lock);

	if (is_bad_inode(inode) || xfs_iflags_test(ip, XFS_INEW)) {
		IRELE(ip);
		return ENOENT;
	}

	return 0;
}

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
STATIC int
xfs_sync_inode_data(
	struct xfs_inode	*ip,
	int			flags)
{
	struct inode		*inode = VFS_I(ip);
	struct address_space *mapping = inode->i_mapping;
	int			error = 0;

	if (!mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
		goto out_wait;

	if (!xfs_ilock_nowait(ip, XFS_IOLOCK_SHARED)) {
		if (flags & SYNC_TRYLOCK)
			goto out_wait;
		xfs_ilock(ip, XFS_IOLOCK_SHARED);
	}

	error = xfs_flush_pages(ip, 0, -1, (flags & SYNC_WAIT) ?
				0 : XFS_B_ASYNC, FI_NONE);
	xfs_iunlock(ip, XFS_IOLOCK_SHARED);

 out_wait:
	if (flags & SYNC_IOWAIT)
		xfs_ioend_wait(ip);
	return error;
}

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
STATIC int
xfs_sync_inode_attr(
	struct xfs_inode	*ip,
	int			flags)
{
	int			error = 0;

	xfs_ilock(ip, XFS_ILOCK_SHARED);
	if (xfs_inode_clean(ip))
		goto out_unlock;
	if (!xfs_iflock_nowait(ip)) {
		if (!(flags & SYNC_WAIT))
			goto out_unlock;
		xfs_iflock(ip);
	}

	if (xfs_inode_clean(ip)) {
		xfs_ifunlock(ip);
		goto out_unlock;
	}

	error = xfs_iflush(ip, (flags & SYNC_WAIT) ?
			   XFS_IFLUSH_SYNC : XFS_IFLUSH_DELWRI);

 out_unlock:
	xfs_iunlock(ip, XFS_ILOCK_SHARED);
	return error;
}

142
/*
143 144
 * Sync all the inodes in the given AG according to the
 * direction given by the flags.
145
 */
146 147
STATIC int
xfs_sync_inodes_ag(
148
	xfs_mount_t	*mp,
149
	int		ag,
150
	int		flags)
151
{
152 153
	xfs_perag_t	*pag = &mp->m_perag[ag];
	int		nr_found;
154
	uint32_t	first_index = 0;
155 156
	int		error = 0;
	int		last_error = 0;
157 158

	do {
159 160
		xfs_inode_t	*ip = NULL;

161
		/*
162 163 164
		 * use a gang lookup to find the next inode in the tree
		 * as the tree is sparse and a gang lookup walks to find
		 * the number of objects requested.
165
		 */
166 167 168
		read_lock(&pag->pag_ici_lock);
		nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
				(void**)&ip, first_index, 1);
169

170 171 172
		if (!nr_found) {
			read_unlock(&pag->pag_ici_lock);
			break;
173 174
		}

175 176 177 178 179 180
		/*
		 * Update the index for the next lookup. Catch overflows
		 * into the next AG range which can occur if we have inodes
		 * in the last block of the AG and we are currently
		 * pointing to the last inode.
		 */
181
		first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
182 183 184 185
		if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) {
			read_unlock(&pag->pag_ici_lock);
			break;
		}
186

187 188 189 190
		error = xfs_sync_inode_valid(ip, pag);
		if (error) {
			if (error == EFSCORRUPTED)
				return 0;
191
			continue;
192
		}
193

194 195
		/*
		 * If we have to flush data or wait for I/O completion
196
		 * we need to hold the iolock.
197
		 */
198 199
		if (flags & SYNC_DELWRI)
			error = xfs_sync_inode_data(ip, flags);
200

201 202 203 204
		if (flags & SYNC_ATTR)
			error = xfs_sync_inode_attr(ip, flags);

		IRELE(ip);
205

206
		if (error)
207 208 209 210
			last_error = error;
		/*
		 * bail out if the filesystem is corrupted.
		 */
211
		if (error == EFSCORRUPTED)
212 213
			return XFS_ERROR(error);

214
	} while (nr_found);
215

216 217
	return last_error;
}
218

219 220 221
int
xfs_sync_inodes(
	xfs_mount_t	*mp,
222
	int		flags)
223 224 225 226
{
	int		error;
	int		last_error;
	int		i;
227
	int		lflags = XFS_LOG_FORCE;
228

229 230 231 232
	if (mp->m_flags & XFS_MOUNT_RDONLY)
		return 0;
	error = 0;
	last_error = 0;
233

234 235 236
	if (flags & SYNC_WAIT)
		lflags |= XFS_LOG_SYNC;

237 238 239
	for (i = 0; i < mp->m_sb.sb_agcount; i++) {
		if (!mp->m_perag[i].pag_ici_init)
			continue;
240
		error = xfs_sync_inodes_ag(mp, i, flags);
241 242 243 244 245
		if (error)
			last_error = error;
		if (error == EFSCORRUPTED)
			break;
	}
246 247 248
	if (flags & SYNC_DELWRI)
		xfs_log_force(mp, 0, lflags);

249 250 251
	return XFS_ERROR(last_error);
}

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
STATIC int
xfs_commit_dummy_trans(
	struct xfs_mount	*mp,
	uint			log_flags)
{
	struct xfs_inode	*ip = mp->m_rootip;
	struct xfs_trans	*tp;
	int			error;

	/*
	 * Put a dummy transaction in the log to tell recovery
	 * that all others are OK.
	 */
	tp = xfs_trans_alloc(mp, XFS_TRANS_DUMMY1);
	error = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
	if (error) {
		xfs_trans_cancel(tp, 0);
		return error;
	}

	xfs_ilock(ip, XFS_ILOCK_EXCL);

	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
	xfs_trans_ihold(tp, ip);
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
	/* XXX(hch): ignoring the error here.. */
	error = xfs_trans_commit(tp, 0);

	xfs_iunlock(ip, XFS_ILOCK_EXCL);

	xfs_log_force(mp, 0, log_flags);
	return 0;
}

286
int
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
xfs_sync_fsdata(
	struct xfs_mount	*mp,
	int			flags)
{
	struct xfs_buf		*bp;
	struct xfs_buf_log_item	*bip;
	int			error = 0;

	/*
	 * If this is xfssyncd() then only sync the superblock if we can
	 * lock it without sleeping and it is not pinned.
	 */
	if (flags & SYNC_BDFLUSH) {
		ASSERT(!(flags & SYNC_WAIT));

		bp = xfs_getsb(mp, XFS_BUF_TRYLOCK);
		if (!bp)
			goto out;

		bip = XFS_BUF_FSPRIVATE(bp, struct xfs_buf_log_item *);
		if (!bip || !xfs_buf_item_dirty(bip) || XFS_BUF_ISPINNED(bp))
			goto out_brelse;
	} else {
		bp = xfs_getsb(mp, 0);

		/*
		 * If the buffer is pinned then push on the log so we won't
		 * get stuck waiting in the write for someone, maybe
		 * ourselves, to flush the log.
		 *
		 * Even though we just pushed the log above, we did not have
		 * the superblock buffer locked at that point so it can
		 * become pinned in between there and here.
		 */
		if (XFS_BUF_ISPINNED(bp))
			xfs_log_force(mp, 0, XFS_LOG_FORCE);
	}


	if (flags & SYNC_WAIT)
		XFS_BUF_UNASYNC(bp);
	else
		XFS_BUF_ASYNC(bp);

	return xfs_bwrite(mp, bp);

 out_brelse:
	xfs_buf_relse(bp);
 out:
	return error;
337 338 339
}

/*
D
David Chinner 已提交
340 341 342 343 344 345 346 347 348 349 350
 * When remounting a filesystem read-only or freezing the filesystem, we have
 * two phases to execute. This first phase is syncing the data before we
 * quiesce the filesystem, and the second is flushing all the inodes out after
 * we've waited for all the transactions created by the first phase to
 * complete. The second phase ensures that the inodes are written to their
 * location on disk rather than just existing in transactions in the log. This
 * means after a quiesce there is no log replay required to write the inodes to
 * disk (this is the main difference between a sync and a quiesce).
 */
/*
 * First stage of freeze - no writers will make progress now we are here,
351 352
 * so we flush delwri and delalloc buffers here, then wait for all I/O to
 * complete.  Data is frozen at that point. Metadata is not frozen,
D
David Chinner 已提交
353 354
 * transactions can still occur here so don't bother flushing the buftarg
 * because it'll just get dirty again.
355 356 357 358 359 360 361 362 363
 */
int
xfs_quiesce_data(
	struct xfs_mount	*mp)
{
	int error;

	/* push non-blocking */
	xfs_sync_inodes(mp, SYNC_DELWRI|SYNC_BDFLUSH);
C
Christoph Hellwig 已提交
364
	xfs_qm_sync(mp, SYNC_BDFLUSH);
365 366 367 368
	xfs_filestream_flush(mp);

	/* push and block */
	xfs_sync_inodes(mp, SYNC_DELWRI|SYNC_WAIT|SYNC_IOWAIT);
C
Christoph Hellwig 已提交
369
	xfs_qm_sync(mp, SYNC_WAIT);
370

D
David Chinner 已提交
371
	/* write superblock and hoover up shutdown errors */
372 373
	error = xfs_sync_fsdata(mp, 0);

D
David Chinner 已提交
374
	/* flush data-only devices */
375 376 377 378
	if (mp->m_rtdev_targp)
		XFS_bflush(mp->m_rtdev_targp);

	return error;
379 380
}

D
David Chinner 已提交
381 382 383 384 385 386 387
STATIC void
xfs_quiesce_fs(
	struct xfs_mount	*mp)
{
	int	count = 0, pincount;

	xfs_flush_buftarg(mp->m_ddev_targp, 0);
388
	xfs_reclaim_inodes(mp, 0, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
D
David Chinner 已提交
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

	/*
	 * This loop must run at least twice.  The first instance of the loop
	 * will flush most meta data but that will generate more meta data
	 * (typically directory updates).  Which then must be flushed and
	 * logged before we can write the unmount record.
	 */
	do {
		xfs_sync_inodes(mp, SYNC_ATTR|SYNC_WAIT);
		pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
		if (!pincount) {
			delay(50);
			count++;
		}
	} while (count < 2);
}

/*
 * Second stage of a quiesce. The data is already synced, now we have to take
 * care of the metadata. New transactions are already blocked, so we need to
 * wait for any remaining transactions to drain out before proceding.
 */
void
xfs_quiesce_attr(
	struct xfs_mount	*mp)
{
	int	error = 0;

	/* wait for all modifications to complete */
	while (atomic_read(&mp->m_active_trans) > 0)
		delay(100);

	/* flush inodes and push all remaining buffers out to disk */
	xfs_quiesce_fs(mp);

424 425 426 427 428
	/*
	 * Just warn here till VFS can correctly support
	 * read-only remount without racing.
	 */
	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
D
David Chinner 已提交
429 430 431 432 433 434 435 436 437 438 439

	/* Push the superblock and write an unmount record */
	error = xfs_log_sbcount(mp, 1);
	if (error)
		xfs_fs_cmn_err(CE_WARN, mp,
				"xfs_attr_quiesce: failed to log sb changes. "
				"Frozen image may not be consistent.");
	xfs_log_unmount_write(mp);
	xfs_unmountfs_writesb(mp);
}

440 441 442 443 444 445 446 447 448 449 450
/*
 * Enqueue a work item to be picked up by the vfs xfssyncd thread.
 * Doing this has two advantages:
 * - It saves on stack space, which is tight in certain situations
 * - It can be used (with care) as a mechanism to avoid deadlocks.
 * Flushing while allocating in a full filesystem requires both.
 */
STATIC void
xfs_syncd_queue_work(
	struct xfs_mount *mp,
	void		*data,
451 452
	void		(*syncer)(struct xfs_mount *, void *),
	struct completion *completion)
453
{
454
	struct xfs_sync_work *work;
455

456
	work = kmem_alloc(sizeof(struct xfs_sync_work), KM_SLEEP);
457 458 459 460
	INIT_LIST_HEAD(&work->w_list);
	work->w_syncer = syncer;
	work->w_data = data;
	work->w_mount = mp;
461
	work->w_completion = completion;
462 463 464 465 466 467 468 469 470 471 472 473 474
	spin_lock(&mp->m_sync_lock);
	list_add_tail(&work->w_list, &mp->m_sync_list);
	spin_unlock(&mp->m_sync_lock);
	wake_up_process(mp->m_sync_task);
}

/*
 * Flush delayed allocate data, attempting to free up reserved space
 * from existing allocations.  At this point a new allocation attempt
 * has failed with ENOSPC and we are in the process of scratching our
 * heads, looking about for more room...
 */
STATIC void
475
xfs_flush_inodes_work(
476 477 478 479
	struct xfs_mount *mp,
	void		*arg)
{
	struct inode	*inode = arg;
480 481
	xfs_sync_inodes(mp, SYNC_DELWRI | SYNC_TRYLOCK);
	xfs_sync_inodes(mp, SYNC_DELWRI | SYNC_TRYLOCK | SYNC_IOWAIT);
482 483 484 485
	iput(inode);
}

void
486
xfs_flush_inodes(
487 488 489
	xfs_inode_t	*ip)
{
	struct inode	*inode = VFS_I(ip);
490
	DECLARE_COMPLETION_ONSTACK(completion);
491 492

	igrab(inode);
493 494
	xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_inodes_work, &completion);
	wait_for_completion(&completion);
495 496 497
	xfs_log_force(ip->i_mount, (xfs_lsn_t)0, XFS_LOG_FORCE|XFS_LOG_SYNC);
}

498 499 500 501 502
/*
 * Every sync period we need to unpin all items, reclaim inodes, sync
 * quota and write out the superblock. We might need to cover the log
 * to indicate it is idle.
 */
503 504 505 506 507 508 509
STATIC void
xfs_sync_worker(
	struct xfs_mount *mp,
	void		*unused)
{
	int		error;

510 511
	if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
		xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
512
		xfs_reclaim_inodes(mp, 0, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
513
		/* dgc: errors ignored here */
C
Christoph Hellwig 已提交
514
		error = xfs_qm_sync(mp, SYNC_BDFLUSH);
515 516 517 518
		error = xfs_sync_fsdata(mp, SYNC_BDFLUSH);
		if (xfs_log_need_covered(mp))
			error = xfs_commit_dummy_trans(mp, XFS_LOG_FORCE);
	}
519 520 521 522 523 524 525 526 527 528
	mp->m_sync_seq++;
	wake_up(&mp->m_wait_single_sync_task);
}

STATIC int
xfssyncd(
	void			*arg)
{
	struct xfs_mount	*mp = arg;
	long			timeleft;
529
	xfs_sync_work_t		*work, *n;
530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
	LIST_HEAD		(tmp);

	set_freezable();
	timeleft = xfs_syncd_centisecs * msecs_to_jiffies(10);
	for (;;) {
		timeleft = schedule_timeout_interruptible(timeleft);
		/* swsusp */
		try_to_freeze();
		if (kthread_should_stop() && list_empty(&mp->m_sync_list))
			break;

		spin_lock(&mp->m_sync_lock);
		/*
		 * We can get woken by laptop mode, to do a sync -
		 * that's the (only!) case where the list would be
		 * empty with time remaining.
		 */
		if (!timeleft || list_empty(&mp->m_sync_list)) {
			if (!timeleft)
				timeleft = xfs_syncd_centisecs *
							msecs_to_jiffies(10);
			INIT_LIST_HEAD(&mp->m_sync_work.w_list);
			list_add_tail(&mp->m_sync_work.w_list,
					&mp->m_sync_list);
		}
		list_for_each_entry_safe(work, n, &mp->m_sync_list, w_list)
			list_move(&work->w_list, &tmp);
		spin_unlock(&mp->m_sync_lock);

		list_for_each_entry_safe(work, n, &tmp, w_list) {
			(*work->w_syncer)(mp, work->w_data);
			list_del(&work->w_list);
			if (work == &mp->m_sync_work)
				continue;
564 565
			if (work->w_completion)
				complete(work->w_completion);
566 567 568 569 570 571 572 573 574 575 576 577 578
			kmem_free(work);
		}
	}

	return 0;
}

int
xfs_syncd_init(
	struct xfs_mount	*mp)
{
	mp->m_sync_work.w_syncer = xfs_sync_worker;
	mp->m_sync_work.w_mount = mp;
579
	mp->m_sync_work.w_completion = NULL;
580 581 582 583 584 585 586 587 588 589 590 591 592
	mp->m_sync_task = kthread_run(xfssyncd, mp, "xfssyncd");
	if (IS_ERR(mp->m_sync_task))
		return -PTR_ERR(mp->m_sync_task);
	return 0;
}

void
xfs_syncd_stop(
	struct xfs_mount	*mp)
{
	kthread_stop(mp->m_sync_task);
}

593
int
594
xfs_reclaim_inode(
595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653
	xfs_inode_t	*ip,
	int		locked,
	int		sync_mode)
{
	xfs_perag_t	*pag = xfs_get_perag(ip->i_mount, ip->i_ino);

	/* The hash lock here protects a thread in xfs_iget_core from
	 * racing with us on linking the inode back with a vnode.
	 * Once we have the XFS_IRECLAIM flag set it will not touch
	 * us.
	 */
	write_lock(&pag->pag_ici_lock);
	spin_lock(&ip->i_flags_lock);
	if (__xfs_iflags_test(ip, XFS_IRECLAIM) ||
	    !__xfs_iflags_test(ip, XFS_IRECLAIMABLE)) {
		spin_unlock(&ip->i_flags_lock);
		write_unlock(&pag->pag_ici_lock);
		if (locked) {
			xfs_ifunlock(ip);
			xfs_iunlock(ip, XFS_ILOCK_EXCL);
		}
		return 1;
	}
	__xfs_iflags_set(ip, XFS_IRECLAIM);
	spin_unlock(&ip->i_flags_lock);
	write_unlock(&pag->pag_ici_lock);
	xfs_put_perag(ip->i_mount, pag);

	/*
	 * If the inode is still dirty, then flush it out.  If the inode
	 * is not in the AIL, then it will be OK to flush it delwri as
	 * long as xfs_iflush() does not keep any references to the inode.
	 * We leave that decision up to xfs_iflush() since it has the
	 * knowledge of whether it's OK to simply do a delwri flush of
	 * the inode or whether we need to wait until the inode is
	 * pulled from the AIL.
	 * We get the flush lock regardless, though, just to make sure
	 * we don't free it while it is being flushed.
	 */
	if (!locked) {
		xfs_ilock(ip, XFS_ILOCK_EXCL);
		xfs_iflock(ip);
	}

	/*
	 * In the case of a forced shutdown we rely on xfs_iflush() to
	 * wait for the inode to be unpinned before returning an error.
	 */
	if (!is_bad_inode(VFS_I(ip)) && xfs_iflush(ip, sync_mode) == 0) {
		/* synchronize with xfs_iflush_done */
		xfs_iflock(ip);
		xfs_ifunlock(ip);
	}

	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	xfs_ireclaim(ip);
	return 0;
}

D
David Chinner 已提交
654 655 656 657 658
/*
 * We set the inode flag atomically with the radix tree tag.
 * Once we get tag lookups on the radix tree, this inode flag
 * can go away.
 */
659 660 661 662 663 664 665 666 667 668 669
void
xfs_inode_set_reclaim_tag(
	xfs_inode_t	*ip)
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_perag_t	*pag = xfs_get_perag(mp, ip->i_ino);

	read_lock(&pag->pag_ici_lock);
	spin_lock(&ip->i_flags_lock);
	radix_tree_tag_set(&pag->pag_ici_root,
			XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
D
David Chinner 已提交
670
	__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
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
	spin_unlock(&ip->i_flags_lock);
	read_unlock(&pag->pag_ici_lock);
	xfs_put_perag(mp, pag);
}

void
__xfs_inode_clear_reclaim_tag(
	xfs_mount_t	*mp,
	xfs_perag_t	*pag,
	xfs_inode_t	*ip)
{
	radix_tree_tag_clear(&pag->pag_ici_root,
			XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
}

void
xfs_inode_clear_reclaim_tag(
	xfs_inode_t	*ip)
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_perag_t	*pag = xfs_get_perag(mp, ip->i_ino);

	read_lock(&pag->pag_ici_lock);
	spin_lock(&ip->i_flags_lock);
	__xfs_inode_clear_reclaim_tag(mp, pag, ip);
	spin_unlock(&ip->i_flags_lock);
	read_unlock(&pag->pag_ici_lock);
	xfs_put_perag(mp, pag);
}

701 702 703

STATIC void
xfs_reclaim_inodes_ag(
704
	xfs_mount_t	*mp,
705 706
	int		ag,
	int		noblock,
707 708
	int		mode)
{
709 710 711
	xfs_inode_t	*ip = NULL;
	xfs_perag_t	*pag = &mp->m_perag[ag];
	int		nr_found;
712
	uint32_t	first_index;
713
	int		skipped;
714 715

restart:
716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
	first_index = 0;
	skipped = 0;
	do {
		/*
		 * use a gang lookup to find the next inode in the tree
		 * as the tree is sparse and a gang lookup walks to find
		 * the number of objects requested.
		 */
		read_lock(&pag->pag_ici_lock);
		nr_found = radix_tree_gang_lookup_tag(&pag->pag_ici_root,
					(void**)&ip, first_index, 1,
					XFS_ICI_RECLAIM_TAG);

		if (!nr_found) {
			read_unlock(&pag->pag_ici_lock);
			break;
		}

734 735 736 737 738 739
		/*
		 * Update the index for the next lookup. Catch overflows
		 * into the next AG range which can occur if we have inodes
		 * in the last block of the AG and we are currently
		 * pointing to the last inode.
		 */
740
		first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
741 742 743 744
		if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) {
			read_unlock(&pag->pag_ici_lock);
			break;
		}
745 746 747 748 749 750 751

		/* ignore if already under reclaim */
		if (xfs_iflags_test(ip, XFS_IRECLAIM)) {
			read_unlock(&pag->pag_ici_lock);
			continue;
		}

752
		if (noblock) {
753 754
			if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
				read_unlock(&pag->pag_ici_lock);
755
				continue;
756
			}
757 758 759
			if (xfs_ipincount(ip) ||
			    !xfs_iflock_nowait(ip)) {
				xfs_iunlock(ip, XFS_ILOCK_EXCL);
760
				read_unlock(&pag->pag_ici_lock);
761 762 763
				continue;
			}
		}
764 765 766 767 768 769
		read_unlock(&pag->pag_ici_lock);

		/*
		 * hmmm - this is an inode already in reclaim. Do
		 * we even bother catching it here?
		 */
770
		if (xfs_reclaim_inode(ip, noblock, mode))
771 772 773 774 775
			skipped++;
	} while (nr_found);

	if (skipped) {
		delay(1);
776 777
		goto restart;
	}
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794
	return;

}

int
xfs_reclaim_inodes(
	xfs_mount_t	*mp,
	int		 noblock,
	int		mode)
{
	int		i;

	for (i = 0; i < mp->m_sb.sb_agcount; i++) {
		if (!mp->m_perag[i].pag_ici_init)
			continue;
		xfs_reclaim_inodes_ag(mp, i, noblock, mode);
	}
795 796 797 798
	return 0;
}