aboutsummaryrefslogtreecommitdiff
path: root/fs/xfs/xfs_vfsops.c
blob: 8b5a3376c2f71e73133dafa787fe7fd610731dff (plain)
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
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
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
218
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
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
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
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
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
758
759
760
761
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
837
838
839
840
841
842
843
844
845
846
847
848
849
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
/*
 * 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_da_btree.h"
#include "xfs_bmap_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_inode_item.h"
#include "xfs_btree.h"
#include "xfs_alloc.h"
#include "xfs_ialloc.h"
#include "xfs_quota.h"
#include "xfs_error.h"
#include "xfs_bmap.h"
#include "xfs_rw.h"
#include "xfs_buf_item.h"
#include "xfs_log_priv.h"
#include "xfs_dir2_trace.h"
#include "xfs_extfree_item.h"
#include "xfs_acl.h"
#include "xfs_attr.h"
#include "xfs_clnt.h"
#include "xfs_mru_cache.h"
#include "xfs_filestream.h"
#include "xfs_fsops.h"
#include "xfs_vnodeops.h"
#include "xfs_vfsops.h"
#include "xfs_utils.h"


int __init
xfs_init(void)
{
#ifdef XFS_DABUF_DEBUG
	extern spinlock_t        xfs_dabuf_global_lock;
	spin_lock_init(&xfs_dabuf_global_lock);
#endif

	/*
	 * Initialize all of the zone allocators we use.
	 */
	xfs_log_ticket_zone = kmem_zone_init(sizeof(xlog_ticket_t),
						"xfs_log_ticket");
	xfs_bmap_free_item_zone = kmem_zone_init(sizeof(xfs_bmap_free_item_t),
						"xfs_bmap_free_item");
	xfs_btree_cur_zone = kmem_zone_init(sizeof(xfs_btree_cur_t),
						"xfs_btree_cur");
	xfs_da_state_zone = kmem_zone_init(sizeof(xfs_da_state_t),
						"xfs_da_state");
	xfs_dabuf_zone = kmem_zone_init(sizeof(xfs_dabuf_t), "xfs_dabuf");
	xfs_ifork_zone = kmem_zone_init(sizeof(xfs_ifork_t), "xfs_ifork");
	xfs_trans_zone = kmem_zone_init(sizeof(xfs_trans_t), "xfs_trans");
	xfs_acl_zone_init(xfs_acl_zone, "xfs_acl");
	xfs_mru_cache_init();
	xfs_filestream_init();

	/*
	 * The size of the zone allocated buf log item is the maximum
	 * size possible under XFS.  This wastes a little bit of memory,
	 * but it is much faster.
	 */
	xfs_buf_item_zone =
		kmem_zone_init((sizeof(xfs_buf_log_item_t) +
				(((XFS_MAX_BLOCKSIZE / XFS_BLI_CHUNK) /
				  NBWORD) * sizeof(int))),
			       "xfs_buf_item");
	xfs_efd_zone =
		kmem_zone_init((sizeof(xfs_efd_log_item_t) +
			       ((XFS_EFD_MAX_FAST_EXTENTS - 1) *
				 sizeof(xfs_extent_t))),
				      "xfs_efd_item");
	xfs_efi_zone =
		kmem_zone_init((sizeof(xfs_efi_log_item_t) +
			       ((XFS_EFI_MAX_FAST_EXTENTS - 1) *
				 sizeof(xfs_extent_t))),
				      "xfs_efi_item");

	/*
	 * These zones warrant special memory allocator hints
	 */
	xfs_inode_zone =
		kmem_zone_init_flags(sizeof(xfs_inode_t), "xfs_inode",
					KM_ZONE_HWALIGN | KM_ZONE_RECLAIM |
					KM_ZONE_SPREAD, NULL);
	xfs_ili_zone =
		kmem_zone_init_flags(sizeof(xfs_inode_log_item_t), "xfs_ili",
					KM_ZONE_SPREAD, NULL);

	/*
	 * Allocate global trace buffers.
	 */
#ifdef XFS_ALLOC_TRACE
	xfs_alloc_trace_buf = ktrace_alloc(XFS_ALLOC_TRACE_SIZE, KM_SLEEP);
#endif
#ifdef XFS_BMAP_TRACE
	xfs_bmap_trace_buf = ktrace_alloc(XFS_BMAP_TRACE_SIZE, KM_SLEEP);
#endif
#ifdef XFS_BMBT_TRACE
	xfs_bmbt_trace_buf = ktrace_alloc(XFS_BMBT_TRACE_SIZE, KM_SLEEP);
#endif
#ifdef XFS_ATTR_TRACE
	xfs_attr_trace_buf = ktrace_alloc(XFS_ATTR_TRACE_SIZE, KM_SLEEP);
#endif
#ifdef XFS_DIR2_TRACE
	xfs_dir2_trace_buf = ktrace_alloc(XFS_DIR2_GTRACE_SIZE, KM_SLEEP);
#endif

	xfs_dir_startup();

#if (defined(DEBUG) || defined(INDUCE_IO_ERROR))
	xfs_error_test_init();
#endif /* DEBUG || INDUCE_IO_ERROR */

	xfs_init_procfs();
	xfs_sysctl_register();
	return 0;
}

void __exit
xfs_cleanup(void)
{
	extern kmem_zone_t	*xfs_inode_zone;
	extern kmem_zone_t	*xfs_efd_zone;
	extern kmem_zone_t	*xfs_efi_zone;

	xfs_cleanup_procfs();
	xfs_sysctl_unregister();
	xfs_filestream_uninit();
	xfs_mru_cache_uninit();
	xfs_acl_zone_destroy(xfs_acl_zone);

#ifdef XFS_DIR2_TRACE
	ktrace_free(xfs_dir2_trace_buf);
#endif
#ifdef XFS_ATTR_TRACE
	ktrace_free(xfs_attr_trace_buf);
#endif
#ifdef XFS_BMBT_TRACE
	ktrace_free(xfs_bmbt_trace_buf);
#endif
#ifdef XFS_BMAP_TRACE
	ktrace_free(xfs_bmap_trace_buf);
#endif
#ifdef XFS_ALLOC_TRACE
	ktrace_free(xfs_alloc_trace_buf);
#endif

	kmem_zone_destroy(xfs_bmap_free_item_zone);
	kmem_zone_destroy(xfs_btree_cur_zone);
	kmem_zone_destroy(xfs_inode_zone);
	kmem_zone_destroy(xfs_trans_zone);
	kmem_zone_destroy(xfs_da_state_zone);
	kmem_zone_destroy(xfs_dabuf_zone);
	kmem_zone_destroy(xfs_buf_item_zone);
	kmem_zone_destroy(xfs_efd_zone);
	kmem_zone_destroy(xfs_efi_zone);
	kmem_zone_destroy(xfs_ifork_zone);
	kmem_zone_destroy(xfs_ili_zone);
	kmem_zone_destroy(xfs_log_ticket_zone);
}

STATIC void
xfs_quiesce_fs(
	xfs_mount_t		*mp)
{
	int			count = 0, pincount;

	xfs_flush_buftarg(mp->m_ddev_targp, 0);
	xfs_finish_reclaim_all(mp, 0);

	/* 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_syncsub(mp, SYNC_INODE_QUIESCE, NULL);
		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_attr_quiesce(
	xfs_mount_t	*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);

	ASSERT_ALWAYS(atomic_read(&mp->m_active_trans) == 0);

	/* 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);
}

/*
 * xfs_unmount_flush implements a set of flush operation on special
 * inodes, which are needed as a separate set of operations so that
 * they can be called as part of relocation process.
 */
int
xfs_unmount_flush(
	xfs_mount_t	*mp,		/* Mount structure we are getting
					   rid of. */
	int             relocation)	/* Called from vfs relocation. */
{
	xfs_inode_t	*rip = mp->m_rootip;
	xfs_inode_t	*rbmip;
	xfs_inode_t	*rsumip = NULL;
	bhv_vnode_t	*rvp = XFS_ITOV(rip);
	int		error;

	xfs_ilock(rip, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
	xfs_iflock(rip);

	/*
	 * Flush out the real time inodes.
	 */
	if ((rbmip = mp->m_rbmip) != NULL) {
		xfs_ilock(rbmip, XFS_ILOCK_EXCL);
		xfs_iflock(rbmip);
		error = xfs_iflush(rbmip, XFS_IFLUSH_SYNC);
		xfs_iunlock(rbmip, XFS_ILOCK_EXCL);

		if (error == EFSCORRUPTED)
			goto fscorrupt_out;

		ASSERT(vn_count(XFS_ITOV(rbmip)) == 1);

		rsumip = mp->m_rsumip;
		xfs_ilock(rsumip, XFS_ILOCK_EXCL);
		xfs_iflock(rsumip);
		error = xfs_iflush(rsumip, XFS_IFLUSH_SYNC);
		xfs_iunlock(rsumip, XFS_ILOCK_EXCL);

		if (error == EFSCORRUPTED)
			goto fscorrupt_out;

		ASSERT(vn_count(XFS_ITOV(rsumip)) == 1);
	}

	/*
	 * Synchronously flush root inode to disk
	 */
	error = xfs_iflush(rip, XFS_IFLUSH_SYNC);
	if (error == EFSCORRUPTED)
		goto fscorrupt_out2;

	if (vn_count(rvp) != 1 && !relocation) {
		xfs_iunlock(rip, XFS_ILOCK_EXCL);
		return XFS_ERROR(EBUSY);
	}

	/*
	 * Release dquot that rootinode, rbmino and rsumino might be holding,
	 * flush and purge the quota inodes.
	 */
	error = XFS_QM_UNMOUNT(mp);
	if (error == EFSCORRUPTED)
		goto fscorrupt_out2;

	if (rbmip) {
		IRELE(rbmip);
		IRELE(rsumip);
	}

	xfs_iunlock(rip, XFS_ILOCK_EXCL);
	return 0;

fscorrupt_out:
	xfs_ifunlock(rip);

fscorrupt_out2:
	xfs_iunlock(rip, XFS_ILOCK_EXCL);

	return XFS_ERROR(EFSCORRUPTED);
}

/*
 * xfs_sync flushes any pending I/O to file system vfsp.
 *
 * This routine is called by vfs_sync() to make sure that things make it
 * out to disk eventually, on sync() system calls to flush out everything,
 * and when the file system is unmounted.  For the vfs_sync() case, all
 * we really need to do is sync out the log to make all of our meta-data
 * updates permanent (except for timestamps).  For calls from pflushd(),
 * dirty pages are kept moving by calling pdflush() on the inodes
 * containing them.  We also flush the inodes that we can lock without
 * sleeping and the superblock if we can lock it without sleeping from
 * vfs_sync() so that items at the tail of the log are always moving out.
 *
 * Flags:
 *      SYNC_BDFLUSH - We're being called from vfs_sync() so we don't want
 *		       to sleep if we can help it.  All we really need
 *		       to do is ensure that the log is synced at least
 *		       periodically.  We also push the inodes and
 *		       superblock if we can lock them without sleeping
 *			and they are not pinned.
 *      SYNC_ATTR    - We need to flush the inodes.  If SYNC_BDFLUSH is not
 *		       set, then we really want to lock each inode and flush
 *		       it.
 *      SYNC_WAIT    - All the flushes that take place in this call should
 *		       be synchronous.
 *      SYNC_DELWRI  - This tells us to push dirty pages associated with
 *		       inodes.  SYNC_WAIT and SYNC_BDFLUSH are used to
 *		       determine if they should be flushed sync, async, or
 *		       delwri.
 *      SYNC_CLOSE   - This flag is passed when the system is being
 *		       unmounted.  We should sync and invalidate everything.
 *      SYNC_FSDATA  - This indicates that the caller would like to make
 *		       sure the superblock is safe on disk.  We can ensure
 *		       this by simply making sure the log gets flushed
 *		       if SYNC_BDFLUSH is set, and by actually writing it
 *		       out otherwise.
 *	SYNC_IOWAIT  - The caller wants us to wait for all data I/O to complete
 *		       before we return (including direct I/O). Forms the drain
 *		       side of the write barrier needed to safely quiesce the
 *		       filesystem.
 *
 */
int
xfs_sync(
	xfs_mount_t	*mp,
	int		flags)
{
	int		error;

	/*
	 * Get the Quota Manager to flush the dquots.
	 *
	 * If XFS quota support is not enabled or this filesystem
	 * instance does not use quotas XFS_QM_DQSYNC will always
	 * return zero.
	 */
	error = XFS_QM_DQSYNC(mp, flags);
	if (error) {
		/*
		 * If we got an IO error, we will be shutting down.
		 * So, there's nothing more for us to do here.
		 */
		ASSERT(error != EIO || XFS_FORCED_SHUTDOWN(mp));
		if (XFS_FORCED_SHUTDOWN(mp))
			return XFS_ERROR(error);
	}

	if (flags & SYNC_IOWAIT)
		xfs_filestream_flush(mp);

	return xfs_syncsub(mp, flags, NULL);
}

/*
 * xfs sync routine for internal use
 *
 * This routine supports all of the flags defined for the generic vfs_sync
 * interface as explained above under xfs_sync.
 *
 */
int
xfs_sync_inodes(
	xfs_mount_t	*mp,
	int		flags,
	int             *bypassed)
{
	xfs_inode_t	*ip = NULL;
	bhv_vnode_t	*vp = NULL;
	int		error;
	int		last_error;
	uint64_t	fflag;
	uint		lock_flags;
	uint		base_lock_flags;
	boolean_t	mount_locked;
	boolean_t	vnode_refed;
	int		preempt;
	xfs_iptr_t	*ipointer;
#ifdef DEBUG
	boolean_t	ipointer_in = B_FALSE;

#define IPOINTER_SET	ipointer_in = B_TRUE
#define IPOINTER_CLR	ipointer_in = B_FALSE
#else
#define IPOINTER_SET
#define IPOINTER_CLR
#endif


/* Insert a marker record into the inode list after inode ip. The list
 * must be locked when this is called. After the call the list will no
 * longer be locked.
 */
#define IPOINTER_INSERT(ip, mp)	{ \
		ASSERT(ipointer_in == B_FALSE); \
		ipointer->ip_mnext = ip->i_mnext; \
		ipointer->ip_mprev = ip; \
		ip->i_mnext = (xfs_inode_t *)ipointer; \
		ipointer->ip_mnext->i_mprev = (xfs_inode_t *)ipointer; \
		preempt = 0; \
		XFS_MOUNT_IUNLOCK(mp); \
		mount_locked = B_FALSE; \
		IPOINTER_SET; \
	}

/* Remove the marker from the inode list. If the marker was the only item
 * in the list then there are no remaining inodes and we should zero out
 * the whole list. If we are the current head of the list then move the head
 * past us.
 */
#define IPOINTER_REMOVE(ip, mp)	{ \
		ASSERT(ipointer_in == B_TRUE); \
		if (ipointer->ip_mnext != (xfs_inode_t *)ipointer) { \
			ip = ipointer->ip_mnext; \
			ip->i_mprev = ipointer->ip_mprev; \
			ipointer->ip_mprev->i_mnext = ip; \
			if (mp->m_inodes == (xfs_inode_t *)ipointer) { \
				mp->m_inodes = ip; \
			} \
		} else { \
			ASSERT(mp->m_inodes == (xfs_inode_t *)ipointer); \
			mp->m_inodes = NULL; \
			ip = NULL; \
		} \
		IPOINTER_CLR; \
	}

#define XFS_PREEMPT_MASK	0x7f

	ASSERT(!(flags & SYNC_BDFLUSH));

	if (bypassed)
		*bypassed = 0;
	if (mp->m_flags & XFS_MOUNT_RDONLY)
		return 0;
	error = 0;
	last_error = 0;
	preempt = 0;

	/* Allocate a reference marker */
	ipointer = (xfs_iptr_t *)kmem_zalloc(sizeof(xfs_iptr_t), KM_SLEEP);

	fflag = XFS_B_ASYNC;		/* default is don't wait */
	if (flags & SYNC_DELWRI)
		fflag = XFS_B_DELWRI;
	if (flags & SYNC_WAIT)
		fflag = 0;		/* synchronous overrides all */

	base_lock_flags = XFS_ILOCK_SHARED;
	if (flags & (SYNC_DELWRI | SYNC_CLOSE)) {
		/*
		 * We need the I/O lock if we're going to call any of
		 * the flush/inval routines.
		 */
		base_lock_flags |= XFS_IOLOCK_SHARED;
	}

	XFS_MOUNT_ILOCK(mp);

	ip = mp->m_inodes;

	mount_locked = B_TRUE;
	vnode_refed  = B_FALSE;

	IPOINTER_CLR;

	do {
		ASSERT(ipointer_in == B_FALSE);
		ASSERT(vnode_refed == B_FALSE);

		lock_flags = base_lock_flags;

		/*
		 * There were no inodes in the list, just break out
		 * of the loop.
		 */
		if (ip == NULL) {
			break;
		}

		/*
		 * We found another sync thread marker - skip it
		 */
		if (ip->i_mount == NULL) {
			ip = ip->i_mnext;
			continue;
		}

		vp = XFS_ITOV_NULL(ip);

		/*
		 * If the vnode is gone then this is being torn down,
		 * call reclaim if it is flushed, else let regular flush
		 * code deal with it later in the loop.
		 */

		if (vp == NULL) {
			/* Skip ones already in reclaim */
			if (ip->i_flags & XFS_IRECLAIM) {
				ip = ip->i_mnext;
				continue;
			}
			if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL) == 0) {
				ip = ip->i_mnext;
			} else if ((xfs_ipincount(ip) == 0) &&
				    xfs_iflock_nowait(ip)) {
				IPOINTER_INSERT(ip, mp);

				xfs_finish_reclaim(ip, 1,
						XFS_IFLUSH_DELWRI_ELSE_ASYNC);

				XFS_MOUNT_ILOCK(mp);
				mount_locked = B_TRUE;
				IPOINTER_REMOVE(ip, mp);
			} else {
				xfs_iunlock(ip, XFS_ILOCK_EXCL);
				ip = ip->i_mnext;
			}
			continue;
		}

		if (VN_BAD(vp)) {
			ip = ip->i_mnext;
			continue;
		}

		if (XFS_FORCED_SHUTDOWN(mp) && !(flags & SYNC_CLOSE)) {
			XFS_MOUNT_IUNLOCK(mp);
			kmem_free(ipointer);
			return 0;
		}

		/*
		 * Try to lock without sleeping.  We're out of order with
		 * the inode list lock here, so if we fail we need to drop
		 * the mount lock and try again.  If we're called from
		 * bdflush() here, then don't bother.
		 *
		 * The inode lock here actually coordinates with the
		 * almost spurious inode lock in xfs_ireclaim() to prevent
		 * the vnode we handle here without a reference from
		 * being freed while we reference it.  If we lock the inode
		 * while it's on the mount list here, then the spurious inode
		 * lock in xfs_ireclaim() after the inode is pulled from
		 * the mount list will sleep until we release it here.
		 * This keeps the vnode from being freed while we reference
		 * it.
		 */
		if (xfs_ilock_nowait(ip, lock_flags) == 0) {
			if (vp == NULL) {
				ip = ip->i_mnext;
				continue;
			}

			vp = vn_grab(vp);
			if (vp == NULL) {
				ip = ip->i_mnext;
				continue;
			}

			IPOINTER_INSERT(ip, mp);
			xfs_ilock(ip, lock_flags);

			ASSERT(vp == XFS_ITOV(ip));
			ASSERT(ip->i_mount == mp);

			vnode_refed = B_TRUE;
		}

		/* From here on in the loop we may have a marker record
		 * in the inode list.
		 */

		/*
		 * If we have to flush data or wait for I/O completion
		 * we need to drop the ilock that we currently hold.
		 * If we need to drop the lock, insert a marker if we
		 * have not already done so.
		 */
		if ((flags & (SYNC_CLOSE|SYNC_IOWAIT)) ||
		    ((flags & SYNC_DELWRI) && VN_DIRTY(vp))) {
			if (mount_locked) {
				IPOINTER_INSERT(ip, mp);
			}
			xfs_iunlock(ip, XFS_ILOCK_SHARED);

			if (flags & SYNC_CLOSE) {
				/* Shutdown case. Flush and invalidate. */
				if (XFS_FORCED_SHUTDOWN(mp))
					xfs_tosspages(ip, 0, -1,
							     FI_REMAPF);
				else
					error = xfs_flushinval_pages(ip,
							0, -1, FI_REMAPF);
			} else if ((flags & SYNC_DELWRI) && VN_DIRTY(vp)) {
				error = xfs_flush_pages(ip, 0,
							-1, fflag, FI_NONE);
			}

			/*
			 * When freezing, we need to wait ensure all I/O (including direct
			 * I/O) is complete to ensure no further data modification can take
			 * place after this point
			 */
			if (flags & SYNC_IOWAIT)
				vn_iowait(ip);

			xfs_ilock(ip, XFS_ILOCK_SHARED);
		}

		if ((flags & SYNC_ATTR) &&
		    (ip->i_update_core ||
		     (ip->i_itemp && ip->i_itemp->ili_format.ilf_fields))) {
			if (mount_locked)
				IPOINTER_INSERT(ip, mp);

			if (flags & SYNC_WAIT) {
				xfs_iflock(ip);
				error = xfs_iflush(ip, XFS_IFLUSH_SYNC);

			/*
			 * If we can't acquire the flush lock, then the inode
			 * is already being flushed so don't bother waiting.
			 *
			 * If we can lock it then do a delwri flush so we can
			 * combine multiple inode flushes in each disk write.
			 */
			} else if (xfs_iflock_nowait(ip)) {
				error = xfs_iflush(ip, XFS_IFLUSH_DELWRI);
			} else if (bypassed) {
				(*bypassed)++;
			}
		}

		if (lock_flags != 0) {
			xfs_iunlock(ip, lock_flags);
		}

		if (vnode_refed) {
			/*
			 * If we had to take a reference on the vnode
			 * above, then wait until after we've unlocked
			 * the inode to release the reference.  This is
			 * because we can be already holding the inode
			 * lock when IRELE() calls xfs_inactive().
			 *
			 * Make sure to drop the mount lock before calling
			 * IRELE() so that we don't trip over ourselves if
			 * we have to go for the mount lock again in the
			 * inactive code.
			 */
			if (mount_locked) {
				IPOINTER_INSERT(ip, mp);
			}

			IRELE(ip);

			vnode_refed = B_FALSE;
		}

		if (error) {
			last_error = error;
		}

		/*
		 * bail out if the filesystem is corrupted.
		 */
		if (error == EFSCORRUPTED)  {
			if (!mount_locked) {
				XFS_MOUNT_ILOCK(mp);
				IPOINTER_REMOVE(ip, mp);
			}
			XFS_MOUNT_IUNLOCK(mp);
			ASSERT(ipointer_in == B_FALSE);
			kmem_free(ipointer);
			return XFS_ERROR(error);
		}

		/* Let other threads have a chance at the mount lock
		 * if we have looped many times without dropping the
		 * lock.
		 */
		if ((++preempt & XFS_PREEMPT_MASK) == 0) {
			if (mount_locked) {
				IPOINTER_INSERT(ip, mp);
			}
		}

		if (mount_locked == B_FALSE) {
			XFS_MOUNT_ILOCK(mp);
			mount_locked = B_TRUE;
			IPOINTER_REMOVE(ip, mp);
			continue;
		}

		ASSERT(ipointer_in == B_FALSE);
		ip = ip->i_mnext;

	} while (ip != mp->m_inodes);

	XFS_MOUNT_IUNLOCK(mp);

	ASSERT(ipointer_in == B_FALSE);

	kmem_free(ipointer);
	return XFS_ERROR(last_error);
}

/*
 * xfs sync routine for internal use
 *
 * This routine supports all of the flags defined for the generic vfs_sync
 * interface as explained above under xfs_sync.
 *
 */
int
xfs_syncsub(
	xfs_mount_t	*mp,
	int		flags,
	int             *bypassed)
{
	int		error = 0;
	int		last_error = 0;
	uint		log_flags = XFS_LOG_FORCE;
	xfs_buf_t	*bp;
	xfs_buf_log_item_t	*bip;

	/*
	 * Sync out the log.  This ensures that the log is periodically
	 * flushed even if there is not enough activity to fill it up.
	 */
	if (flags & SYNC_WAIT)
		log_flags |= XFS_LOG_SYNC;

	xfs_log_force(mp, (xfs_lsn_t)0, log_flags);

	if (flags & (SYNC_ATTR|SYNC_DELWRI)) {
		if (flags & SYNC_BDFLUSH)
			xfs_finish_reclaim_all(mp, 1);
		else
			error = xfs_sync_inodes(mp, flags, bypassed);
	}

	/*
	 * Flushing out dirty data above probably generated more
	 * log activity, so if this isn't vfs_sync() then flush
	 * the log again.
	 */
	if (flags & SYNC_DELWRI) {
		xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
	}

	if (flags & SYNC_FSDATA) {
		/*
		 * If this is vfs_sync() then only sync the superblock
		 * if we can lock it without sleeping and it is not pinned.
		 */
		if (flags & SYNC_BDFLUSH) {
			bp = xfs_getsb(mp, XFS_BUF_TRYLOCK);
			if (bp != NULL) {
				bip = XFS_BUF_FSPRIVATE(bp,xfs_buf_log_item_t*);
				if ((bip != NULL) &&
				    xfs_buf_item_dirty(bip)) {
					if (!(XFS_BUF_ISPINNED(bp))) {
						XFS_BUF_ASYNC(bp);
						error = xfs_bwrite(mp, bp);
					} else {
						xfs_buf_relse(bp);
					}
				} else {
					xfs_buf_relse(bp);
				}
			}
		} 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, (xfs_lsn_t)0, XFS_LOG_FORCE);
			if (flags & SYNC_WAIT)
				XFS_BUF_UNASYNC(bp);
			else
				XFS_BUF_ASYNC(bp);
			error = xfs_bwrite(mp, bp);
		}
		if (error) {
			last_error = error;
		}
	}

	/*
	 * Now check to see if the log needs a "dummy" transaction.
	 */
	if (!(flags & SYNC_REMOUNT) && xfs_log_need_covered(mp)) {
		xfs_trans_t *tp;
		xfs_inode_t *ip;

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

		ip = mp->m_rootip;
		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);
		error = xfs_trans_commit(tp, 0);
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
		xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
	}

	/*
	 * When shutting down, we need to insure that the AIL is pushed
	 * to disk or the filesystem can appear corrupt from the PROM.
	 */
	if ((flags & (SYNC_CLOSE|SYNC_WAIT)) == (SYNC_CLOSE|SYNC_WAIT)) {
		XFS_bflush(mp->m_ddev_targp);
		if (mp->m_rtdev_targp) {
			XFS_bflush(mp->m_rtdev_targp);
		}
	}

	return XFS_ERROR(last_error);
}