module/zfs/vdev_root.c

   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 https://opensource.org/licenses/CDDL-1.0.
  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, 2016 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
  40 static uint64_t
  41 vdev_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  */
  66 static boolean_t
  67 too_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
  83 static int
  84 vdev_root_open(vdev_t *vd, uint64_t *asize, uint64_t *max_asize,
  85     uint64_t *ashift, uint64_t *pshift)
  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                     cvd->vdev_ops != &vdev_indirect_ops) {
 103                         lasterror = cvd->vdev_open_error;
 104                         numerrors++;
 105                 }
 106         }
 107
 108         if (spa_load_state(spa) != SPA_LOAD_NONE)
 109                 spa_set_missing_tvds(spa, numerrors);
 110
 111         if (too_many_errors(vd, numerrors)) {
 112                 vd->vdev_stat.vs_aux = VDEV_AUX_NO_REPLICAS;
 113                 return (lasterror);
 114         }
 115
 116         *asize = 0;
 117         *max_asize = 0;
 118         *ashift = 0;
 119         *pshift = 0;
 120
 121         return (0);
 122 }
 123
 124 static void
 125 vdev_root_close(vdev_t *vd)
 126 {
 127         for (int c = 0; c < vd->vdev_children; c++)
 128                 vdev_close(vd->vdev_child[c]);
 129 }
 130
 131 static void
 132 vdev_root_state_change(vdev_t *vd, int faulted, int degraded)
 133 {
 134         if (too_many_errors(vd, faulted)) {
 135                 vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
 136                     VDEV_AUX_NO_REPLICAS);
 137         } else if (degraded || faulted) {
 138                 vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE);
 139         } else {
 140                 vdev_set_state(vd, B_FALSE, VDEV_STATE_HEALTHY, VDEV_AUX_NONE);
 141         }
 142 }
 143
 144 vdev_ops_t vdev_root_ops = {
 145         .vdev_op_init = NULL,
 146         .vdev_op_fini = NULL,
 147         .vdev_op_open = vdev_root_open,
 148         .vdev_op_close = vdev_root_close,
 149         .vdev_op_asize = vdev_default_asize,
 150         .vdev_op_min_asize = vdev_default_min_asize,
 151         .vdev_op_min_alloc = NULL,
 152         .vdev_op_io_start = NULL,       /* not applicable to the root */
 153         .vdev_op_io_done = NULL,        /* not applicable to the root */
 154         .vdev_op_state_change = vdev_root_state_change,
 155         .vdev_op_need_resilver = NULL,
 156         .vdev_op_hold = NULL,
 157         .vdev_op_rele = NULL,
 158         .vdev_op_remap = NULL,
 159         .vdev_op_xlate = NULL,
 160         .vdev_op_rebuild_asize = NULL,
 161         .vdev_op_metaslab_init = NULL,
 162         .vdev_op_config_generate = NULL,
 163         .vdev_op_nparity = NULL,
 164         .vdev_op_ndisks = NULL,
 165         .vdev_op_type = VDEV_TYPE_ROOT, /* name of this vdev type */
 166         .vdev_op_leaf = B_FALSE         /* not a leaf vdev */
 167 };