module/zfs/ddt_zap.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 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 /*
  23  * Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
  24  * Copyright (c) 2018 by Delphix. All rights reserved.
  25  */
  26
  27 #include <sys/zfs_context.h>
  28 #include <sys/spa.h>
  29 #include <sys/zio.h>
  30 #include <sys/ddt.h>
  31 #include <sys/zap.h>
  32 #include <sys/dmu_tx.h>
  33
  34 int ddt_zap_leaf_blockshift = 12;
  35 int ddt_zap_indirect_blockshift = 12;
  36
  37 static int
  38 ddt_zap_create(objset_t *os, uint64_t *objectp, dmu_tx_t *tx, boolean_t prehash)
  39 {
  40         zap_flags_t flags = ZAP_FLAG_HASH64 | ZAP_FLAG_UINT64_KEY;
  41
  42         if (prehash)
  43                 flags |= ZAP_FLAG_PRE_HASHED_KEY;
  44
  45         *objectp = zap_create_flags(os, 0, flags, DMU_OT_DDT_ZAP,
  46             ddt_zap_leaf_blockshift, ddt_zap_indirect_blockshift,
  47             DMU_OT_NONE, 0, tx);
  48
  49         return (*objectp == 0 ? SET_ERROR(ENOTSUP) : 0);
  50 }
  51
  52 static int
  53 ddt_zap_destroy(objset_t *os, uint64_t object, dmu_tx_t *tx)
  54 {
  55         return (zap_destroy(os, object, tx));
  56 }
  57
  58 static int
  59 ddt_zap_lookup(objset_t *os, uint64_t object, ddt_entry_t *dde)
  60 {
  61         uchar_t *cbuf;
  62         uint64_t one, csize;
  63         int error;
  64
  65         cbuf = kmem_alloc(sizeof (dde->dde_phys) + 1, KM_SLEEP);
  66
  67         error = zap_length_uint64(os, object, (uint64_t *)&dde->dde_key,
  68             DDT_KEY_WORDS, &one, &csize);
  69         if (error)
  70                 goto out;
  71
  72         ASSERT(one == 1);
  73         ASSERT(csize <= (sizeof (dde->dde_phys) + 1));
  74
  75         error = zap_lookup_uint64(os, object, (uint64_t *)&dde->dde_key,
  76             DDT_KEY_WORDS, 1, csize, cbuf);
  77         if (error)
  78                 goto out;
  79
  80         ddt_decompress(cbuf, dde->dde_phys, csize, sizeof (dde->dde_phys));
  81 out:
  82         kmem_free(cbuf, sizeof (dde->dde_phys) + 1);
  83
  84         return (error);
  85 }
  86
  87 static void
  88 ddt_zap_prefetch(objset_t *os, uint64_t object, ddt_entry_t *dde)
  89 {
  90         (void) zap_prefetch_uint64(os, object, (uint64_t *)&dde->dde_key,
  91             DDT_KEY_WORDS);
  92 }
  93
  94 static int
  95 ddt_zap_update(objset_t *os, uint64_t object, ddt_entry_t *dde, dmu_tx_t *tx)
  96 {
  97         uchar_t cbuf[sizeof (dde->dde_phys) + 1];
  98         uint64_t csize;
  99
 100         csize = ddt_compress(dde->dde_phys, cbuf,
 101             sizeof (dde->dde_phys), sizeof (cbuf));
 102
 103         return (zap_update_uint64(os, object, (uint64_t *)&dde->dde_key,
 104             DDT_KEY_WORDS, 1, csize, cbuf, tx));
 105 }
 106
 107 static int
 108 ddt_zap_remove(objset_t *os, uint64_t object, ddt_entry_t *dde, dmu_tx_t *tx)
 109 {
 110         return (zap_remove_uint64(os, object, (uint64_t *)&dde->dde_key,
 111             DDT_KEY_WORDS, tx));
 112 }
 113
 114 static int
 115 ddt_zap_walk(objset_t *os, uint64_t object, ddt_entry_t *dde, uint64_t *walk)
 116 {
 117         zap_cursor_t zc;
 118         zap_attribute_t za;
 119         int error;
 120
 121         if (*walk == 0) {
 122                 /*
 123                  * We don't want to prefetch the entire ZAP object, because
 124                  * it can be enormous.  Also the primary use of DDT iteration
 125                  * is for scrubbing, in which case we will be issuing many
 126                  * scrub I/Os for each ZAP block that we read in, so
 127                  * reading the ZAP is unlikely to be the bottleneck.
 128                  */
 129                 zap_cursor_init_noprefetch(&zc, os, object);
 130         } else {
 131                 zap_cursor_init_serialized(&zc, os, object, *walk);
 132         }
 133         if ((error = zap_cursor_retrieve(&zc, &za)) == 0) {
 134                 uchar_t cbuf[sizeof (dde->dde_phys) + 1];
 135                 uint64_t csize = za.za_num_integers;
 136                 ASSERT(za.za_integer_length == 1);
 137                 error = zap_lookup_uint64(os, object, (uint64_t *)za.za_name,
 138                     DDT_KEY_WORDS, 1, csize, cbuf);
 139                 ASSERT(error == 0);
 140                 if (error == 0) {
 141                         ddt_decompress(cbuf, dde->dde_phys, csize,
 142                             sizeof (dde->dde_phys));
 143                         dde->dde_key = *(ddt_key_t *)za.za_name;
 144                 }
 145                 zap_cursor_advance(&zc);
 146                 *walk = zap_cursor_serialize(&zc);
 147         }
 148         zap_cursor_fini(&zc);
 149         return (error);
 150 }
 151
 152 static int
 153 ddt_zap_count(objset_t *os, uint64_t object, uint64_t *count)
 154 {
 155         return (zap_count(os, object, count));
 156 }
 157
 158 const ddt_ops_t ddt_zap_ops = {
 159         "zap",
 160         ddt_zap_create,
 161         ddt_zap_destroy,
 162         ddt_zap_lookup,
 163         ddt_zap_prefetch,
 164         ddt_zap_update,
 165         ddt_zap_remove,
 166         ddt_zap_walk,
 167         ddt_zap_count,
 168 };