vdev_root.c revision 332536 
1/*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21/*
22 * Copyright 2010 Sun Microsystems, Inc.  All rights reserved.
23 * Use is subject to license terms.
24 */
25
26/*
27 * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
28 */
29
30#include <sys/zfs_context.h>
31#include <sys/spa.h>
32#include <sys/vdev_impl.h>
33#include <sys/zio.h>
34#include <sys/fs/zfs.h>
35
36/*
37 * Virtual device vector for the pool's root vdev.
38 */
39
40static uint64_t
41vdev_root_core_tvds(vdev_t *vd)
42{
43	uint64_t tvds = 0;
44
45	for (uint64_t c = 0; c < vd->vdev_children; c++) {
46		vdev_t *cvd = vd->vdev_child[c];
47
48		if (!cvd->vdev_ishole && !cvd->vdev_islog &&
49		    cvd->vdev_ops != &vdev_indirect_ops) {
50			tvds++;
51		}
52	}
53
54	return (tvds);
55}
56
57/*
58 * We should be able to tolerate one failure with absolutely no damage
59 * to our metadata.  Two failures will take out space maps, a bunch of
60 * indirect block trees, meta dnodes, dnodes, etc.  Probably not a happy
61 * place to live.  When we get smarter, we can liberalize this policy.
62 * e.g. If we haven't lost two consecutive top-level vdevs, then we are
63 * probably fine.  Adding bean counters during alloc/free can make this
64 * future guesswork more accurate.
65 */
66static boolean_t
67too_many_errors(vdev_t *vd, uint64_t numerrors)
68{
69	uint64_t tvds;
70
71	if (numerrors == 0)
72		return (B_FALSE);
73
74	tvds = vdev_root_core_tvds(vd);
75	ASSERT3U(numerrors, <=, tvds);
76
77	if (numerrors == tvds)
78		return (B_TRUE);
79
80	return (numerrors > spa_missing_tvds_allowed(vd->vdev_spa));
81}
82
83static int
84vdev_root_open(vdev_t *vd, uint64_t *asize, uint64_t *max_asize,
85    uint64_t *logical_ashift, uint64_t *physical_ashift)
86{
87	spa_t *spa = vd->vdev_spa;
88	int lasterror = 0;
89	int numerrors = 0;
90
91	if (vd->vdev_children == 0) {
92		vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
93		return (SET_ERROR(EINVAL));
94	}
95
96	vdev_open_children(vd);
97
98	for (int c = 0; c < vd->vdev_children; c++) {
99		vdev_t *cvd = vd->vdev_child[c];
100
101		if (cvd->vdev_open_error && !cvd->vdev_islog) {
102			lasterror = cvd->vdev_open_error;
103			numerrors++;
104		}
105	}
106
107	if (spa_load_state(spa) != SPA_LOAD_NONE)
108		spa_set_missing_tvds(spa, numerrors);
109
110	if (too_many_errors(vd, numerrors)) {
111		vd->vdev_stat.vs_aux = VDEV_AUX_NO_REPLICAS;
112		return (lasterror);
113	}
114
115	*asize = 0;
116	*max_asize = 0;
117	*logical_ashift = 0;
118	*physical_ashift = 0;
119
120	return (0);
121}
122
123static void
124vdev_root_close(vdev_t *vd)
125{
126	for (int c = 0; c < vd->vdev_children; c++)
127		vdev_close(vd->vdev_child[c]);
128}
129
130static void
131vdev_root_state_change(vdev_t *vd, int faulted, int degraded)
132{
133	if (too_many_errors(vd, faulted)) {
134		vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
135		    VDEV_AUX_NO_REPLICAS);
136	} else if (degraded || faulted) {
137		vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE);
138	} else {
139		vdev_set_state(vd, B_FALSE, VDEV_STATE_HEALTHY, VDEV_AUX_NONE);
140	}
141}
142
143vdev_ops_t vdev_root_ops = {
144	vdev_root_open,
145	vdev_root_close,
146	vdev_default_asize,
147	NULL,			/* io_start - not applicable to the root */
148	NULL,			/* io_done - not applicable to the root */
149	vdev_root_state_change,
150	NULL,
151	NULL,
152	NULL,
153	VDEV_TYPE_ROOT,		/* name of this vdev type */
154	B_FALSE			/* not a leaf vdev */
155};
156