aboutsummaryrefslogtreecommitdiffstats
path: root/uts/common/fs/zfs/vdev_disk.c
blob: 2d431373ce1d6276ae39b661880164c7befe8dc5 (plain) (blame)
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
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
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
974
975
976
977
978
/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
 * Copyright 2016 Nexenta Systems, Inc.  All rights reserved.
 * Copyright (c) 2013 Joyent, Inc.  All rights reserved.
 */

#include <sys/zfs_context.h>
#include <sys/spa_impl.h>
#include <sys/refcount.h>
#include <sys/vdev_disk.h>
#include <sys/vdev_impl.h>
#include <sys/abd.h>
#include <sys/fs/zfs.h>
#include <sys/zio.h>
#include <sys/sunldi.h>
#include <sys/efi_partition.h>
#include <sys/fm/fs/zfs.h>

/*
 * Tunable parameter for debugging or performance analysis. Setting this
 * will cause pool corruption on power loss if a volatile out-of-order
 * write cache is enabled.
 */
boolean_t zfs_nocacheflush = B_FALSE;

/*
 * Virtual device vector for disks.
 */

extern ldi_ident_t zfs_li;

static void vdev_disk_close(vdev_t *);

typedef struct vdev_disk_ldi_cb {
	list_node_t		lcb_next;
	ldi_callback_id_t	lcb_id;
} vdev_disk_ldi_cb_t;

/*
 * Bypass the devid when opening a disk vdev.
 * There have been issues where the devids of several devices were shuffled,
 * causing pool open failures. Note, that this flag is intended to be used
 * for pool recovery only.
 *
 * Note that if a pool is imported with the devids bypassed, all its vdevs will
 * cease storing devid information permanently. In practice, the devid is rarely
 * useful as vdev paths do not tend to change unless the hardware is
 * reconfigured. That said, if the paths do change and a pool fails to open
 * automatically at boot, a simple zpool import should re-scan the paths and fix
 * the issue.
 */
boolean_t vdev_disk_bypass_devid = B_FALSE;

static void
vdev_disk_alloc(vdev_t *vd)
{
	vdev_disk_t *dvd;

	dvd = vd->vdev_tsd = kmem_zalloc(sizeof (vdev_disk_t), KM_SLEEP);
	/*
	 * Create the LDI event callback list.
	 */
	list_create(&dvd->vd_ldi_cbs, sizeof (vdev_disk_ldi_cb_t),
	    offsetof(vdev_disk_ldi_cb_t, lcb_next));
}

static void
vdev_disk_free(vdev_t *vd)
{
	vdev_disk_t *dvd = vd->vdev_tsd;
	vdev_disk_ldi_cb_t *lcb;

	if (dvd == NULL)
		return;

	/*
	 * We have already closed the LDI handle. Clean up the LDI event
	 * callbacks and free vd->vdev_tsd.
	 */
	while ((lcb = list_head(&dvd->vd_ldi_cbs)) != NULL) {
		list_remove(&dvd->vd_ldi_cbs, lcb);
		(void) ldi_ev_remove_callbacks(lcb->lcb_id);
		kmem_free(lcb, sizeof (vdev_disk_ldi_cb_t));
	}
	list_destroy(&dvd->vd_ldi_cbs);
	kmem_free(dvd, sizeof (vdev_disk_t));
	vd->vdev_tsd = NULL;
}

/* ARGSUSED */
static int
vdev_disk_off_notify(ldi_handle_t lh, ldi_ev_cookie_t ecookie, void *arg,
    void *ev_data)
{
	vdev_t *vd = (vdev_t *)arg;
	vdev_disk_t *dvd = vd->vdev_tsd;

	/*
	 * Ignore events other than offline.
	 */
	if (strcmp(ldi_ev_get_type(ecookie), LDI_EV_OFFLINE) != 0)
		return (LDI_EV_SUCCESS);

	/*
	 * All LDI handles must be closed for the state change to succeed, so
	 * call on vdev_disk_close() to do this.
	 *
	 * We inform vdev_disk_close that it is being called from offline
	 * notify context so it will defer cleanup of LDI event callbacks and
	 * freeing of vd->vdev_tsd to the offline finalize or a reopen.
	 */
	dvd->vd_ldi_offline = B_TRUE;
	vdev_disk_close(vd);

	/*
	 * Now that the device is closed, request that the spa_async_thread
	 * mark the device as REMOVED and notify FMA of the removal.
	 */
	zfs_post_remove(vd->vdev_spa, vd);
	vd->vdev_remove_wanted = B_TRUE;
	spa_async_request(vd->vdev_spa, SPA_ASYNC_REMOVE);

	return (LDI_EV_SUCCESS);
}

/* ARGSUSED */
static void
vdev_disk_off_finalize(ldi_handle_t lh, ldi_ev_cookie_t ecookie,
    int ldi_result, void *arg, void *ev_data)
{
	vdev_t *vd = (vdev_t *)arg;

	/*
	 * Ignore events other than offline.
	 */
	if (strcmp(ldi_ev_get_type(ecookie), LDI_EV_OFFLINE) != 0)
		return;

	/*
	 * We have already closed the LDI handle in notify.
	 * Clean up the LDI event callbacks and free vd->vdev_tsd.
	 */
	vdev_disk_free(vd);

	/*
	 * Request that the vdev be reopened if the offline state change was
	 * unsuccessful.
	 */
	if (ldi_result != LDI_EV_SUCCESS) {
		vd->vdev_probe_wanted = B_TRUE;
		spa_async_request(vd->vdev_spa, SPA_ASYNC_PROBE);
	}
}

static ldi_ev_callback_t vdev_disk_off_callb = {
	.cb_vers = LDI_EV_CB_VERS,
	.cb_notify = vdev_disk_off_notify,
	.cb_finalize = vdev_disk_off_finalize
};

/* ARGSUSED */
static void
vdev_disk_dgrd_finalize(ldi_handle_t lh, ldi_ev_cookie_t ecookie,
    int ldi_result, void *arg, void *ev_data)
{
	vdev_t *vd = (vdev_t *)arg;

	/*
	 * Ignore events other than degrade.
	 */
	if (strcmp(ldi_ev_get_type(ecookie), LDI_EV_DEGRADE) != 0)
		return;

	/*
	 * Degrade events always succeed. Mark the vdev as degraded.
	 * This status is purely informative for the user.
	 */
	(void) vdev_degrade(vd->vdev_spa, vd->vdev_guid, 0);
}

static ldi_ev_callback_t vdev_disk_dgrd_callb = {
	.cb_vers = LDI_EV_CB_VERS,
	.cb_notify = NULL,
	.cb_finalize = vdev_disk_dgrd_finalize
};

static void
vdev_disk_hold(vdev_t *vd)
{
	ddi_devid_t devid;
	char *minor;

	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));

	/*
	 * We must have a pathname, and it must be absolute.
	 */
	if (vd->vdev_path == NULL || vd->vdev_path[0] != '/')
		return;

	/*
	 * Only prefetch path and devid info if the device has
	 * never been opened.
	 */
	if (vd->vdev_tsd != NULL)
		return;

	if (vd->vdev_wholedisk == -1ULL) {
		size_t len = strlen(vd->vdev_path) + 3;
		char *buf = kmem_alloc(len, KM_SLEEP);

		(void) snprintf(buf, len, "%ss0", vd->vdev_path);

		(void) ldi_vp_from_name(buf, &vd->vdev_name_vp);
		kmem_free(buf, len);
	}

	if (vd->vdev_name_vp == NULL)
		(void) ldi_vp_from_name(vd->vdev_path, &vd->vdev_name_vp);

	if (vd->vdev_devid != NULL &&
	    ddi_devid_str_decode(vd->vdev_devid, &devid, &minor) == 0) {
		(void) ldi_vp_from_devid(devid, minor, &vd->vdev_devid_vp);
		ddi_devid_str_free(minor);
		ddi_devid_free(devid);
	}
}

static void
vdev_disk_rele(vdev_t *vd)
{
	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));

	if (vd->vdev_name_vp) {
		VN_RELE_ASYNC(vd->vdev_name_vp,
		    dsl_pool_vnrele_taskq(vd->vdev_spa->spa_dsl_pool));
		vd->vdev_name_vp = NULL;
	}
	if (vd->vdev_devid_vp) {
		VN_RELE_ASYNC(vd->vdev_devid_vp,
		    dsl_pool_vnrele_taskq(vd->vdev_spa->spa_dsl_pool));
		vd->vdev_devid_vp = NULL;
	}
}

/*
 * We want to be loud in DEBUG kernels when DKIOCGMEDIAINFOEXT fails, or when
 * even a fallback to DKIOCGMEDIAINFO fails.
 */
#ifdef DEBUG
#define	VDEV_DEBUG(...)	cmn_err(CE_NOTE, __VA_ARGS__)
#else
#define	VDEV_DEBUG(...)	/* Nothing... */
#endif

static int
vdev_disk_open(vdev_t *vd, uint64_t *psize, uint64_t *max_psize,
    uint64_t *ashift)
{
	spa_t *spa = vd->vdev_spa;
	vdev_disk_t *dvd = vd->vdev_tsd;
	ldi_ev_cookie_t ecookie;
	vdev_disk_ldi_cb_t *lcb;
	union {
		struct dk_minfo_ext ude;
		struct dk_minfo ud;
	} dks;
	struct dk_minfo_ext *dkmext = &dks.ude;
	struct dk_minfo *dkm = &dks.ud;
	int error;
	dev_t dev;
	int otyp;
	boolean_t validate_devid = B_FALSE;
	ddi_devid_t devid;
	uint64_t capacity = 0, blksz = 0, pbsize;

	/*
	 * We must have a pathname, and it must be absolute.
	 */
	if (vd->vdev_path == NULL || vd->vdev_path[0] != '/') {
		vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
		return (SET_ERROR(EINVAL));
	}

	/*
	 * Reopen the device if it's not currently open. Otherwise,
	 * just update the physical size of the device.
	 */
	if (dvd != NULL) {
		if (dvd->vd_ldi_offline && dvd->vd_lh == NULL) {
			/*
			 * If we are opening a device in its offline notify
			 * context, the LDI handle was just closed. Clean
			 * up the LDI event callbacks and free vd->vdev_tsd.
			 */
			vdev_disk_free(vd);
		} else {
			ASSERT(vd->vdev_reopening);
			goto skip_open;
		}
	}

	/*
	 * Create vd->vdev_tsd.
	 */
	vdev_disk_alloc(vd);
	dvd = vd->vdev_tsd;

	/*
	 * Allow bypassing the devid.
	 */
	if (vd->vdev_devid != NULL && vdev_disk_bypass_devid) {
		vdev_dbgmsg(vd, "vdev_disk_open, devid %s bypassed",
		    vd->vdev_devid);
		spa_strfree(vd->vdev_devid);
		vd->vdev_devid = NULL;
	}

	/*
	 * When opening a disk device, we want to preserve the user's original
	 * intent.  We always want to open the device by the path the user gave
	 * us, even if it is one of multiple paths to the same device.  But we
	 * also want to be able to survive disks being removed/recabled.
	 * Therefore the sequence of opening devices is:
	 *
	 * 1. Try opening the device by path.  For legacy pools without the
	 *    'whole_disk' property, attempt to fix the path by appending 's0'.
	 *
	 * 2. If the devid of the device matches the stored value, return
	 *    success.
	 *
	 * 3. Otherwise, the device may have moved.  Try opening the device
	 *    by the devid instead.
	 */
	if (vd->vdev_devid != NULL) {
		if (ddi_devid_str_decode(vd->vdev_devid, &dvd->vd_devid,
		    &dvd->vd_minor) != 0) {
			vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
			vdev_dbgmsg(vd, "vdev_disk_open: invalid "
			    "vdev_devid '%s'", vd->vdev_devid);
			return (SET_ERROR(EINVAL));
		}
	}

	error = EINVAL;		/* presume failure */

	if (vd->vdev_path != NULL) {

		if (vd->vdev_wholedisk == -1ULL) {
			size_t len = strlen(vd->vdev_path) + 3;
			char *buf = kmem_alloc(len, KM_SLEEP);

			(void) snprintf(buf, len, "%ss0", vd->vdev_path);

			error = ldi_open_by_name(buf, spa_mode(spa), kcred,
			    &dvd->vd_lh, zfs_li);
			if (error == 0) {
				spa_strfree(vd->vdev_path);
				vd->vdev_path = buf;
				vd->vdev_wholedisk = 1ULL;
			} else {
				kmem_free(buf, len);
			}
		}

		/*
		 * If we have not yet opened the device, try to open it by the
		 * specified path.
		 */
		if (error != 0) {
			error = ldi_open_by_name(vd->vdev_path, spa_mode(spa),
			    kcred, &dvd->vd_lh, zfs_li);
		}

		/*
		 * Compare the devid to the stored value.
		 */
		if (error == 0 && vd->vdev_devid != NULL &&
		    ldi_get_devid(dvd->vd_lh, &devid) == 0) {
			if (ddi_devid_compare(devid, dvd->vd_devid) != 0) {
				/*
				 * A mismatch here is unexpected, log it.
				 */
				char *devid_str = ddi_devid_str_encode(devid,
				    dvd->vd_minor);
				vdev_dbgmsg(vd, "vdev_disk_open: devid "
				    "mismatch: %s != %s", vd->vdev_devid,
				    devid_str);
				cmn_err(CE_NOTE, "vdev_disk_open %s: devid "
				    "mismatch: %s != %s", vd->vdev_path,
				    vd->vdev_devid, devid_str);
				ddi_devid_str_free(devid_str);

				error = SET_ERROR(EINVAL);
				(void) ldi_close(dvd->vd_lh, spa_mode(spa),
				    kcred);
				dvd->vd_lh = NULL;
			}
			ddi_devid_free(devid);
		}

		/*
		 * If we succeeded in opening the device, but 'vdev_wholedisk'
		 * is not yet set, then this must be a slice.
		 */
		if (error == 0 && vd->vdev_wholedisk == -1ULL)
			vd->vdev_wholedisk = 0;
	}

	/*
	 * If we were unable to open by path, or the devid check fails, open by
	 * devid instead.
	 */
	if (error != 0 && vd->vdev_devid != NULL) {
		error = ldi_open_by_devid(dvd->vd_devid, dvd->vd_minor,
		    spa_mode(spa), kcred, &dvd->vd_lh, zfs_li);
		if (error != 0) {
			vdev_dbgmsg(vd, "Failed to open by devid (%s)",
			    vd->vdev_devid);
		}
	}

	/*
	 * If all else fails, then try opening by physical path (if available)
	 * or the logical path (if we failed due to the devid check).  While not
	 * as reliable as the devid, this will give us something, and the higher
	 * level vdev validation will prevent us from opening the wrong device.
	 */
	if (error) {
		if (vd->vdev_devid != NULL)
			validate_devid = B_TRUE;

		if (vd->vdev_physpath != NULL &&
		    (dev = ddi_pathname_to_dev_t(vd->vdev_physpath)) != NODEV)
			error = ldi_open_by_dev(&dev, OTYP_BLK, spa_mode(spa),
			    kcred, &dvd->vd_lh, zfs_li);

		/*
		 * Note that we don't support the legacy auto-wholedisk support
		 * as above.  This hasn't been used in a very long time and we
		 * don't need to propagate its oddities to this edge condition.
		 */
		if (error && vd->vdev_path != NULL)
			error = ldi_open_by_name(vd->vdev_path, spa_mode(spa),
			    kcred, &dvd->vd_lh, zfs_li);
	}

	if (error) {
		vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
		vdev_dbgmsg(vd, "vdev_disk_open: failed to open [error=%d]",
		    error);
		return (error);
	}

	/*
	 * Now that the device has been successfully opened, update the devid
	 * if necessary.
	 */
	if (validate_devid && spa_writeable(spa) &&
	    ldi_get_devid(dvd->vd_lh, &devid) == 0) {
		if (ddi_devid_compare(devid, dvd->vd_devid) != 0) {
			char *vd_devid;

			vd_devid = ddi_devid_str_encode(devid, dvd->vd_minor);
			vdev_dbgmsg(vd, "vdev_disk_open: update devid from "
			    "'%s' to '%s'", vd->vdev_devid, vd_devid);
			cmn_err(CE_NOTE, "vdev_disk_open %s: update devid "
			    "from '%s' to '%s'", vd->vdev_path != NULL ?
			    vd->vdev_path : "?", vd->vdev_devid, vd_devid);
			spa_strfree(vd->vdev_devid);
			vd->vdev_devid = spa_strdup(vd_devid);
			ddi_devid_str_free(vd_devid);
		}
		ddi_devid_free(devid);
	}

	/*
	 * Once a device is opened, verify that the physical device path (if
	 * available) is up to date.
	 */
	if (ldi_get_dev(dvd->vd_lh, &dev) == 0 &&
	    ldi_get_otyp(dvd->vd_lh, &otyp) == 0) {
		char *physpath, *minorname;

		physpath = kmem_alloc(MAXPATHLEN, KM_SLEEP);
		minorname = NULL;
		if (ddi_dev_pathname(dev, otyp, physpath) == 0 &&
		    ldi_get_minor_name(dvd->vd_lh, &minorname) == 0 &&
		    (vd->vdev_physpath == NULL ||
		    strcmp(vd->vdev_physpath, physpath) != 0)) {
			if (vd->vdev_physpath)
				spa_strfree(vd->vdev_physpath);
			(void) strlcat(physpath, ":", MAXPATHLEN);
			(void) strlcat(physpath, minorname, MAXPATHLEN);
			vd->vdev_physpath = spa_strdup(physpath);
		}
		if (minorname)
			kmem_free(minorname, strlen(minorname) + 1);
		kmem_free(physpath, MAXPATHLEN);
	}

	/*
	 * Register callbacks for the LDI offline event.
	 */
	if (ldi_ev_get_cookie(dvd->vd_lh, LDI_EV_OFFLINE, &ecookie) ==
	    LDI_EV_SUCCESS) {
		lcb = kmem_zalloc(sizeof (vdev_disk_ldi_cb_t), KM_SLEEP);
		list_insert_tail(&dvd->vd_ldi_cbs, lcb);
		(void) ldi_ev_register_callbacks(dvd->vd_lh, ecookie,
		    &vdev_disk_off_callb, (void *) vd, &lcb->lcb_id);
	}

	/*
	 * Register callbacks for the LDI degrade event.
	 */
	if (ldi_ev_get_cookie(dvd->vd_lh, LDI_EV_DEGRADE, &ecookie) ==
	    LDI_EV_SUCCESS) {
		lcb = kmem_zalloc(sizeof (vdev_disk_ldi_cb_t), KM_SLEEP);
		list_insert_tail(&dvd->vd_ldi_cbs, lcb);
		(void) ldi_ev_register_callbacks(dvd->vd_lh, ecookie,
		    &vdev_disk_dgrd_callb, (void *) vd, &lcb->lcb_id);
	}
skip_open:
	/*
	 * Determine the actual size of the device.
	 */
	if (ldi_get_size(dvd->vd_lh, psize) != 0) {
		vd->vdev_stat.vs_aux = VDEV_AUX_OPEN_FAILED;
		vdev_dbgmsg(vd, "vdev_disk_open: failed to get size");
		return (SET_ERROR(EINVAL));
	}

	*max_psize = *psize;

	/*
	 * Determine the device's minimum transfer size.
	 * If the ioctl isn't supported, assume DEV_BSIZE.
	 */
	if ((error = ldi_ioctl(dvd->vd_lh, DKIOCGMEDIAINFOEXT,
	    (intptr_t)dkmext, FKIOCTL, kcred, NULL)) == 0) {
		capacity = dkmext->dki_capacity - 1;
		blksz = dkmext->dki_lbsize;
		pbsize = dkmext->dki_pbsize;
	} else if ((error = ldi_ioctl(dvd->vd_lh, DKIOCGMEDIAINFO,
	    (intptr_t)dkm, FKIOCTL, kcred, NULL)) == 0) {
		VDEV_DEBUG(
		    "vdev_disk_open(\"%s\"): fallback to DKIOCGMEDIAINFO\n",
		    vd->vdev_path);
		capacity = dkm->dki_capacity - 1;
		blksz = dkm->dki_lbsize;
		pbsize = blksz;
	} else {
		VDEV_DEBUG("vdev_disk_open(\"%s\"): "
		    "both DKIOCGMEDIAINFO{,EXT} calls failed, %d\n",
		    vd->vdev_path, error);
		pbsize = DEV_BSIZE;
	}

	*ashift = highbit64(MAX(pbsize, SPA_MINBLOCKSIZE)) - 1;

	if (vd->vdev_wholedisk == 1) {
		int wce = 1;

		if (error == 0) {
			/*
			 * If we have the capability to expand, we'd have
			 * found out via success from DKIOCGMEDIAINFO{,EXT}.
			 * Adjust max_psize upward accordingly since we know
			 * we own the whole disk now.
			 */
			*max_psize = capacity * blksz;
		}

		/*
		 * Since we own the whole disk, try to enable disk write
		 * caching.  We ignore errors because it's OK if we can't do it.
		 */
		(void) ldi_ioctl(dvd->vd_lh, DKIOCSETWCE, (intptr_t)&wce,
		    FKIOCTL, kcred, NULL);
	}

	/*
	 * Clear the nowritecache bit, so that on a vdev_reopen() we will
	 * try again.
	 */
	vd->vdev_nowritecache = B_FALSE;

	return (0);
}

static void
vdev_disk_close(vdev_t *vd)
{
	vdev_disk_t *dvd = vd->vdev_tsd;

	if (vd->vdev_reopening || dvd == NULL)
		return;

	if (dvd->vd_minor != NULL) {
		ddi_devid_str_free(dvd->vd_minor);
		dvd->vd_minor = NULL;
	}

	if (dvd->vd_devid != NULL) {
		ddi_devid_free(dvd->vd_devid);
		dvd->vd_devid = NULL;
	}

	if (dvd->vd_lh != NULL) {
		(void) ldi_close(dvd->vd_lh, spa_mode(vd->vdev_spa), kcred);
		dvd->vd_lh = NULL;
	}

	vd->vdev_delayed_close = B_FALSE;
	/*
	 * If we closed the LDI handle due to an offline notify from LDI,
	 * don't free vd->vdev_tsd or unregister the callbacks here;
	 * the offline finalize callback or a reopen will take care of it.
	 */
	if (dvd->vd_ldi_offline)
		return;

	vdev_disk_free(vd);
}

int
vdev_disk_physio(vdev_t *vd, caddr_t data,
    size_t size, uint64_t offset, int flags, boolean_t isdump)
{
	vdev_disk_t *dvd = vd->vdev_tsd;

	/*
	 * If the vdev is closed, it's likely in the REMOVED or FAULTED state.
	 * Nothing to be done here but return failure.
	 */
	if (dvd == NULL || (dvd->vd_ldi_offline && dvd->vd_lh == NULL))
		return (EIO);

	ASSERT(vd->vdev_ops == &vdev_disk_ops);

	/*
	 * If in the context of an active crash dump, use the ldi_dump(9F)
	 * call instead of ldi_strategy(9F) as usual.
	 */
	if (isdump) {
		ASSERT3P(dvd, !=, NULL);
		return (ldi_dump(dvd->vd_lh, data, lbtodb(offset),
		    lbtodb(size)));
	}

	return (vdev_disk_ldi_physio(dvd->vd_lh, data, size, offset, flags));
}

int
vdev_disk_ldi_physio(ldi_handle_t vd_lh, caddr_t data,
    size_t size, uint64_t offset, int flags)
{
	buf_t *bp;
	int error = 0;

	if (vd_lh == NULL)
		return (SET_ERROR(EINVAL));

	ASSERT(flags & B_READ || flags & B_WRITE);

	bp = getrbuf(KM_SLEEP);
	bp->b_flags = flags | B_BUSY | B_NOCACHE | B_FAILFAST;
	bp->b_bcount = size;
	bp->b_un.b_addr = (void *)data;
	bp->b_lblkno = lbtodb(offset);
	bp->b_bufsize = size;

	error = ldi_strategy(vd_lh, bp);
	ASSERT(error == 0);
	if ((error = biowait(bp)) == 0 && bp->b_resid != 0)
		error = SET_ERROR(EIO);
	freerbuf(bp);

	return (error);
}

static int
vdev_disk_io_intr(buf_t *bp)
{
	vdev_buf_t *vb = (vdev_buf_t *)bp;
	zio_t *zio = vb->vb_io;

	/*
	 * The rest of the zio stack only deals with EIO, ECKSUM, and ENXIO.
	 * Rather than teach the rest of the stack about other error
	 * possibilities (EFAULT, etc), we normalize the error value here.
	 */
	zio->io_error = (geterror(bp) != 0 ? EIO : 0);

	if (zio->io_error == 0 && bp->b_resid != 0)
		zio->io_error = SET_ERROR(EIO);

	if (zio->io_type == ZIO_TYPE_READ) {
		abd_return_buf_copy(zio->io_abd, bp->b_un.b_addr, zio->io_size);
	} else {
		abd_return_buf(zio->io_abd, bp->b_un.b_addr, zio->io_size);
	}

	kmem_free(vb, sizeof (vdev_buf_t));

	zio_delay_interrupt(zio);
	return (0);
}

static void
vdev_disk_ioctl_free(zio_t *zio)
{
	kmem_free(zio->io_vsd, sizeof (struct dk_callback));
}

static const zio_vsd_ops_t vdev_disk_vsd_ops = {
	vdev_disk_ioctl_free,
	zio_vsd_default_cksum_report
};

static void
vdev_disk_ioctl_done(void *zio_arg, int error)
{
	zio_t *zio = zio_arg;

	zio->io_error = error;

	zio_interrupt(zio);
}

static void
vdev_disk_io_start(zio_t *zio)
{
	vdev_t *vd = zio->io_vd;
	vdev_disk_t *dvd = vd->vdev_tsd;
	vdev_buf_t *vb;
	struct dk_callback *dkc;
	buf_t *bp;
	int error;

	/*
	 * If the vdev is closed, it's likely in the REMOVED or FAULTED state.
	 * Nothing to be done here but return failure.
	 */
	if (dvd == NULL || (dvd->vd_ldi_offline && dvd->vd_lh == NULL)) {
		zio->io_error = ENXIO;
		zio_interrupt(zio);
		return;
	}

	if (zio->io_type == ZIO_TYPE_IOCTL) {
		/* XXPOLICY */
		if (!vdev_readable(vd)) {
			zio->io_error = SET_ERROR(ENXIO);
			zio_interrupt(zio);
			return;
		}

		switch (zio->io_cmd) {

		case DKIOCFLUSHWRITECACHE:

			if (zfs_nocacheflush)
				break;

			if (vd->vdev_nowritecache) {
				zio->io_error = SET_ERROR(ENOTSUP);
				break;
			}

			zio->io_vsd = dkc = kmem_alloc(sizeof (*dkc), KM_SLEEP);
			zio->io_vsd_ops = &vdev_disk_vsd_ops;

			dkc->dkc_callback = vdev_disk_ioctl_done;
			dkc->dkc_flag = FLUSH_VOLATILE;
			dkc->dkc_cookie = zio;

			error = ldi_ioctl(dvd->vd_lh, zio->io_cmd,
			    (uintptr_t)dkc, FKIOCTL, kcred, NULL);

			if (error == 0) {
				/*
				 * The ioctl will be done asychronously,
				 * and will call vdev_disk_ioctl_done()
				 * upon completion.
				 */
				return;
			}

			zio->io_error = error;

			break;

		default:
			zio->io_error = SET_ERROR(ENOTSUP);
		}

		zio_execute(zio);
		return;
	}

	ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE);
	zio->io_target_timestamp = zio_handle_io_delay(zio);

	vb = kmem_alloc(sizeof (vdev_buf_t), KM_SLEEP);

	vb->vb_io = zio;
	bp = &vb->vb_buf;

	bioinit(bp);
	bp->b_flags = B_BUSY | B_NOCACHE |
	    (zio->io_type == ZIO_TYPE_READ ? B_READ : B_WRITE);
	if (!(zio->io_flags & (ZIO_FLAG_IO_RETRY | ZIO_FLAG_TRYHARD)))
		bp->b_flags |= B_FAILFAST;
	bp->b_bcount = zio->io_size;

	if (zio->io_type == ZIO_TYPE_READ) {
		bp->b_un.b_addr =
		    abd_borrow_buf(zio->io_abd, zio->io_size);
	} else {
		bp->b_un.b_addr =
		    abd_borrow_buf_copy(zio->io_abd, zio->io_size);
	}

	bp->b_lblkno = lbtodb(zio->io_offset);
	bp->b_bufsize = zio->io_size;
	bp->b_iodone = vdev_disk_io_intr;

	/* ldi_strategy() will return non-zero only on programming errors */
	VERIFY(ldi_strategy(dvd->vd_lh, bp) == 0);
}

static void
vdev_disk_io_done(zio_t *zio)
{
	vdev_t *vd = zio->io_vd;

	/*
	 * If the device returned EIO, then attempt a DKIOCSTATE ioctl to see if
	 * the device has been removed.  If this is the case, then we trigger an
	 * asynchronous removal of the device. Otherwise, probe the device and
	 * make sure it's still accessible.
	 */
	if (zio->io_error == EIO && !vd->vdev_remove_wanted) {
		vdev_disk_t *dvd = vd->vdev_tsd;
		int state = DKIO_NONE;

		if (ldi_ioctl(dvd->vd_lh, DKIOCSTATE, (intptr_t)&state,
		    FKIOCTL, kcred, NULL) == 0 && state != DKIO_INSERTED) {
			/*
			 * We post the resource as soon as possible, instead of
			 * when the async removal actually happens, because the
			 * DE is using this information to discard previous I/O
			 * errors.
			 */
			zfs_post_remove(zio->io_spa, vd);
			vd->vdev_remove_wanted = B_TRUE;
			spa_async_request(zio->io_spa, SPA_ASYNC_REMOVE);
		} else if (!vd->vdev_delayed_close) {
			vd->vdev_delayed_close = B_TRUE;
		}
	}
}

vdev_ops_t vdev_disk_ops = {
	vdev_disk_open,
	vdev_disk_close,
	vdev_default_asize,
	vdev_disk_io_start,
	vdev_disk_io_done,
	NULL,
	vdev_disk_hold,
	vdev_disk_rele,
	NULL,
	vdev_default_xlate,
	VDEV_TYPE_DISK,		/* name of this vdev type */
	B_TRUE			/* leaf vdev */
};

/*
 * Given the root disk device devid or pathname, read the label from
 * the device, and construct a configuration nvlist.
 */
int
vdev_disk_read_rootlabel(char *devpath, char *devid, nvlist_t **config)
{
	ldi_handle_t vd_lh;
	vdev_label_t *label;
	uint64_t s, size;
	int l;
	ddi_devid_t tmpdevid;
	int error = -1;
	char *minor_name;

	/*
	 * Read the device label and build the nvlist.
	 */
	if (devid != NULL && ddi_devid_str_decode(devid, &tmpdevid,
	    &minor_name) == 0) {
		error = ldi_open_by_devid(tmpdevid, minor_name,
		    FREAD, kcred, &vd_lh, zfs_li);
		ddi_devid_free(tmpdevid);
		ddi_devid_str_free(minor_name);
	}

	if (error && (error = ldi_open_by_name(devpath, FREAD, kcred, &vd_lh,
	    zfs_li)))
		return (error);

	if (ldi_get_size(vd_lh, &s)) {
		(void) ldi_close(vd_lh, FREAD, kcred);
		return (SET_ERROR(EIO));
	}

	size = P2ALIGN_TYPED(s, sizeof (vdev_label_t), uint64_t);
	label = kmem_alloc(sizeof (vdev_label_t), KM_SLEEP);

	*config = NULL;
	for (l = 0; l < VDEV_LABELS; l++) {
		uint64_t offset, state, txg = 0;

		/* read vdev label */
		offset = vdev_label_offset(size, l, 0);
		if (vdev_disk_ldi_physio(vd_lh, (caddr_t)label,
		    VDEV_SKIP_SIZE + VDEV_PHYS_SIZE, offset, B_READ) != 0)
			continue;

		if (nvlist_unpack(label->vl_vdev_phys.vp_nvlist,
		    sizeof (label->vl_vdev_phys.vp_nvlist), config, 0) != 0) {
			*config = NULL;
			continue;
		}

		if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
		    &state) != 0 || state >= POOL_STATE_DESTROYED) {
			nvlist_free(*config);
			*config = NULL;
			continue;
		}

		if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
		    &txg) != 0 || txg == 0) {
			nvlist_free(*config);
			*config = NULL;
			continue;
		}

		break;
	}

	kmem_free(label, sizeof (vdev_label_t));
	(void) ldi_close(vd_lh, FREAD, kcred);
	if (*config == NULL)
		error = SET_ERROR(EIDRM);

	return (error);
}