2008-11-20 20:01:55 +00:00
|
|
|
/*
|
|
|
|
* 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
|
|
|
|
*/
|
|
|
|
/*
|
2010-05-28 20:45:14 +00:00
|
|
|
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
|
2014-06-05 21:34:21 +00:00
|
|
|
* Copyright (c) 2012, 2014 by Delphix. All rights reserved.
|
2008-11-20 20:01:55 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/zfs_context.h>
|
|
|
|
#include <sys/dmu_objset.h>
|
|
|
|
#include <sys/dmu_traverse.h>
|
|
|
|
#include <sys/dsl_dataset.h>
|
|
|
|
#include <sys/dsl_dir.h>
|
|
|
|
#include <sys/dsl_pool.h>
|
|
|
|
#include <sys/dnode.h>
|
|
|
|
#include <sys/spa.h>
|
|
|
|
#include <sys/zio.h>
|
|
|
|
#include <sys/dmu_impl.h>
|
2010-05-28 20:45:14 +00:00
|
|
|
#include <sys/sa.h>
|
|
|
|
#include <sys/sa_impl.h>
|
2008-12-03 20:09:06 +00:00
|
|
|
#include <sys/callb.h>
|
2013-12-09 18:37:51 +00:00
|
|
|
#include <sys/zfeature.h>
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2015-03-27 04:31:52 +00:00
|
|
|
int32_t zfs_pd_bytes_max = 50 * 1024 * 1024; /* 50MB */
|
2010-08-26 21:24:34 +00:00
|
|
|
|
|
|
|
typedef struct prefetch_data {
|
2008-12-03 20:09:06 +00:00
|
|
|
kmutex_t pd_mtx;
|
|
|
|
kcondvar_t pd_cv;
|
2015-03-27 04:31:52 +00:00
|
|
|
int32_t pd_bytes_fetched;
|
2008-12-03 20:09:06 +00:00
|
|
|
int pd_flags;
|
|
|
|
boolean_t pd_cancel;
|
|
|
|
boolean_t pd_exited;
|
2010-08-26 21:24:34 +00:00
|
|
|
} prefetch_data_t;
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2010-08-26 21:24:34 +00:00
|
|
|
typedef struct traverse_data {
|
2008-12-03 20:09:06 +00:00
|
|
|
spa_t *td_spa;
|
|
|
|
uint64_t td_objset;
|
|
|
|
blkptr_t *td_rootbp;
|
|
|
|
uint64_t td_min_txg;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t *td_resume;
|
2008-12-03 20:09:06 +00:00
|
|
|
int td_flags;
|
2010-08-26 21:24:34 +00:00
|
|
|
prefetch_data_t *td_pfd;
|
2014-06-05 21:20:08 +00:00
|
|
|
boolean_t td_paused;
|
2008-12-03 20:09:06 +00:00
|
|
|
blkptr_cb_t *td_func;
|
|
|
|
void *td_arg;
|
2010-08-26 21:24:34 +00:00
|
|
|
} traverse_data_t;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2010-08-26 21:24:34 +00:00
|
|
|
static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
|
2013-07-02 20:26:24 +00:00
|
|
|
uint64_t objset, uint64_t object);
|
2013-07-02 20:20:02 +00:00
|
|
|
static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *,
|
2013-07-02 20:26:24 +00:00
|
|
|
uint64_t objset, uint64_t object);
|
2009-07-02 22:44:48 +00:00
|
|
|
|
2010-05-28 20:45:14 +00:00
|
|
|
static int
|
2008-11-20 20:01:55 +00:00
|
|
|
traverse_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg)
|
|
|
|
{
|
2010-08-26 21:24:34 +00:00
|
|
|
traverse_data_t *td = arg;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t zb;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2013-12-09 18:37:51 +00:00
|
|
|
if (BP_IS_HOLE(bp))
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(td->td_spa))
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
|
|
|
|
|
|
|
SET_BOOKMARK(&zb, td->td_objset, ZB_ZIL_OBJECT, ZB_ZIL_LEVEL,
|
|
|
|
bp->blk_cksum.zc_word[ZIL_ZC_SEQ]);
|
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2010-05-28 20:45:14 +00:00
|
|
|
static int
|
2008-11-20 20:01:55 +00:00
|
|
|
traverse_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg)
|
|
|
|
{
|
2010-08-26 21:24:34 +00:00
|
|
|
traverse_data_t *td = arg;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
|
|
|
if (lrc->lrc_txtype == TX_WRITE) {
|
|
|
|
lr_write_t *lr = (lr_write_t *)lrc;
|
|
|
|
blkptr_t *bp = &lr->lr_blkptr;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t zb;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2013-12-09 18:37:51 +00:00
|
|
|
if (BP_IS_HOLE(bp))
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
if (claim_txg == 0 || bp->blk_birth < claim_txg)
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2010-08-26 21:24:34 +00:00
|
|
|
SET_BOOKMARK(&zb, td->td_objset, lr->lr_foid,
|
|
|
|
ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp));
|
2010-05-28 20:45:14 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL,
|
2010-05-28 20:45:14 +00:00
|
|
|
td->td_arg);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
2010-05-28 20:45:14 +00:00
|
|
|
return (0);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2010-08-26 21:24:34 +00:00
|
|
|
traverse_zil(traverse_data_t *td, zil_header_t *zh)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
|
|
|
uint64_t claim_txg = zh->zh_claim_txg;
|
|
|
|
zilog_t *zilog;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We only want to visit blocks that have been claimed but not yet
|
2010-05-28 20:45:14 +00:00
|
|
|
* replayed; plus, in read-only mode, blocks that are already stable.
|
2008-11-20 20:01:55 +00:00
|
|
|
*/
|
2009-01-15 21:59:39 +00:00
|
|
|
if (claim_txg == 0 && spa_writeable(td->td_spa))
|
2008-11-20 20:01:55 +00:00
|
|
|
return;
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
zilog = zil_alloc(spa_get_dsl(td->td_spa)->dp_meta_objset, zh);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
(void) zil_parse(zilog, traverse_zil_block, traverse_zil_record, td,
|
2008-11-20 20:01:55 +00:00
|
|
|
claim_txg);
|
|
|
|
|
|
|
|
zil_free(zilog);
|
|
|
|
}
|
|
|
|
|
2012-12-13 23:24:15 +00:00
|
|
|
typedef enum resume_skip {
|
|
|
|
RESUME_SKIP_ALL,
|
|
|
|
RESUME_SKIP_NONE,
|
|
|
|
RESUME_SKIP_CHILDREN
|
|
|
|
} resume_skip_t;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns RESUME_SKIP_ALL if td indicates that we are resuming a traversal and
|
|
|
|
* the block indicated by zb does not need to be visited at all. Returns
|
|
|
|
* RESUME_SKIP_CHILDREN if we are resuming a post traversal and we reach the
|
|
|
|
* resume point. This indicates that this block should be visited but not its
|
|
|
|
* children (since they must have been visited in a previous traversal).
|
|
|
|
* Otherwise returns RESUME_SKIP_NONE.
|
|
|
|
*/
|
|
|
|
static resume_skip_t
|
|
|
|
resume_skip_check(traverse_data_t *td, const dnode_phys_t *dnp,
|
2014-06-25 18:37:59 +00:00
|
|
|
const zbookmark_phys_t *zb)
|
2012-12-13 23:24:15 +00:00
|
|
|
{
|
|
|
|
if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume)) {
|
|
|
|
/*
|
|
|
|
* If we already visited this bp & everything below,
|
|
|
|
* don't bother doing it again.
|
|
|
|
*/
|
|
|
|
if (zbookmark_is_before(dnp, zb, td->td_resume))
|
|
|
|
return (RESUME_SKIP_ALL);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we found the block we're trying to resume from, zero
|
|
|
|
* the bookmark out to indicate that we have resumed.
|
|
|
|
*/
|
|
|
|
if (bcmp(zb, td->td_resume, sizeof (*zb)) == 0) {
|
|
|
|
bzero(td->td_resume, sizeof (*zb));
|
|
|
|
if (td->td_flags & TRAVERSE_POST)
|
|
|
|
return (RESUME_SKIP_CHILDREN);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return (RESUME_SKIP_NONE);
|
|
|
|
}
|
|
|
|
|
2013-07-02 20:20:02 +00:00
|
|
|
static void
|
|
|
|
traverse_prefetch_metadata(traverse_data_t *td,
|
2014-06-25 18:37:59 +00:00
|
|
|
const blkptr_t *bp, const zbookmark_phys_t *zb)
|
2013-07-02 20:20:02 +00:00
|
|
|
{
|
|
|
|
uint32_t flags = ARC_NOWAIT | ARC_PREFETCH;
|
|
|
|
|
|
|
|
if (!(td->td_flags & TRAVERSE_PREFETCH_METADATA))
|
|
|
|
return;
|
|
|
|
/*
|
|
|
|
* If we are in the process of resuming, don't prefetch, because
|
|
|
|
* some children will not be needed (and in fact may have already
|
|
|
|
* been freed).
|
|
|
|
*/
|
|
|
|
if (td->td_resume != NULL && !ZB_IS_ZERO(td->td_resume))
|
|
|
|
return;
|
|
|
|
if (BP_IS_HOLE(bp) || bp->blk_birth <= td->td_min_txg)
|
|
|
|
return;
|
|
|
|
if (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE)
|
|
|
|
return;
|
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
(void) arc_read(NULL, td->td_spa, bp, NULL, NULL,
|
|
|
|
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
|
2013-07-02 20:20:02 +00:00
|
|
|
}
|
|
|
|
|
2014-06-05 21:34:21 +00:00
|
|
|
static boolean_t
|
|
|
|
prefetch_needed(prefetch_data_t *pfd, const blkptr_t *bp)
|
|
|
|
{
|
|
|
|
ASSERT(pfd->pd_flags & TRAVERSE_PREFETCH_DATA);
|
|
|
|
if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp) ||
|
|
|
|
BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG)
|
|
|
|
return (B_FALSE);
|
|
|
|
return (B_TRUE);
|
|
|
|
}
|
|
|
|
|
2011-05-25 23:09:57 +00:00
|
|
|
static int
|
|
|
|
traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
|
2014-06-25 18:37:59 +00:00
|
|
|
const blkptr_t *bp, const zbookmark_phys_t *zb)
|
2010-08-27 23:48:18 +00:00
|
|
|
{
|
2014-06-05 21:20:08 +00:00
|
|
|
int err = 0;
|
2011-05-25 23:09:57 +00:00
|
|
|
arc_buf_t *buf = NULL;
|
2015-03-27 04:04:12 +00:00
|
|
|
prefetch_data_t *pd = td->td_pfd;
|
2012-12-13 23:24:15 +00:00
|
|
|
|
|
|
|
switch (resume_skip_check(td, dnp, zb)) {
|
|
|
|
case RESUME_SKIP_ALL:
|
|
|
|
return (0);
|
|
|
|
case RESUME_SKIP_CHILDREN:
|
|
|
|
goto post;
|
|
|
|
case RESUME_SKIP_NONE:
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
ASSERT(0);
|
|
|
|
}
|
2010-08-27 23:48:18 +00:00
|
|
|
|
2013-12-09 18:37:51 +00:00
|
|
|
if (bp->blk_birth == 0) {
|
|
|
|
if (spa_feature_is_active(td->td_spa, SPA_FEATURE_HOLE_BIRTH)) {
|
|
|
|
/*
|
|
|
|
* Since this block has a birth time of 0 it must be a
|
|
|
|
* hole created before the SPA_FEATURE_HOLE_BIRTH
|
|
|
|
* feature was enabled. If SPA_FEATURE_HOLE_BIRTH
|
|
|
|
* was enabled before the min_txg for this traveral we
|
|
|
|
* know the hole must have been created before the
|
|
|
|
* min_txg for this traveral, so we can skip it. If
|
|
|
|
* SPA_FEATURE_HOLE_BIRTH was enabled after the min_txg
|
|
|
|
* for this traveral we cannot tell if the hole was
|
|
|
|
* created before or after the min_txg for this
|
|
|
|
* traversal, so we cannot skip it.
|
|
|
|
*/
|
|
|
|
uint64_t hole_birth_enabled_txg;
|
|
|
|
VERIFY(spa_feature_enabled_txg(td->td_spa,
|
|
|
|
SPA_FEATURE_HOLE_BIRTH, &hole_birth_enabled_txg));
|
|
|
|
if (hole_birth_enabled_txg < td->td_min_txg)
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
} else if (bp->blk_birth <= td->td_min_txg) {
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2015-03-27 04:04:12 +00:00
|
|
|
if (pd != NULL && !pd->pd_exited && prefetch_needed(pd, bp)) {
|
2015-03-27 04:31:52 +00:00
|
|
|
uint64_t size = BP_GET_LSIZE(bp);
|
2015-03-27 04:04:12 +00:00
|
|
|
mutex_enter(&pd->pd_mtx);
|
2015-03-27 04:31:52 +00:00
|
|
|
ASSERT(pd->pd_bytes_fetched >= 0);
|
|
|
|
while (pd->pd_bytes_fetched < size && !pd->pd_exited)
|
2015-03-27 04:04:12 +00:00
|
|
|
cv_wait(&pd->pd_cv, &pd->pd_mtx);
|
2015-03-27 04:31:52 +00:00
|
|
|
pd->pd_bytes_fetched -= size;
|
2015-03-27 04:04:12 +00:00
|
|
|
cv_broadcast(&pd->pd_cv);
|
|
|
|
mutex_exit(&pd->pd_mtx);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2014-06-05 21:34:21 +00:00
|
|
|
if (BP_IS_HOLE(bp)) {
|
|
|
|
err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg);
|
|
|
|
if (err != 0)
|
|
|
|
goto post;
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2011-05-25 23:09:57 +00:00
|
|
|
if (td->td_flags & TRAVERSE_PRE) {
|
2013-07-02 20:26:24 +00:00
|
|
|
err = td->td_func(td->td_spa, NULL, bp, zb, dnp,
|
2011-05-25 23:09:57 +00:00
|
|
|
td->td_arg);
|
|
|
|
if (err == TRAVERSE_VISIT_NO_CHILDREN)
|
2010-08-26 21:24:34 +00:00
|
|
|
return (0);
|
2012-12-13 23:24:15 +00:00
|
|
|
if (err != 0)
|
|
|
|
goto post;
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2011-05-25 23:09:57 +00:00
|
|
|
if (BP_GET_LEVEL(bp) > 0) {
|
|
|
|
uint32_t flags = ARC_WAIT;
|
2013-11-13 19:05:17 +00:00
|
|
|
int32_t i;
|
|
|
|
int32_t epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t *czb;
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
|
2011-05-25 23:09:57 +00:00
|
|
|
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0)
|
2014-06-05 21:20:08 +00:00
|
|
|
goto post;
|
2013-11-13 19:05:17 +00:00
|
|
|
|
2014-11-21 00:09:39 +00:00
|
|
|
czb = kmem_alloc(sizeof (zbookmark_phys_t), KM_SLEEP);
|
2013-07-02 20:20:02 +00:00
|
|
|
|
|
|
|
for (i = 0; i < epb; i++) {
|
2013-11-13 19:05:17 +00:00
|
|
|
SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object,
|
2013-07-02 20:20:02 +00:00
|
|
|
zb->zb_level - 1,
|
|
|
|
zb->zb_blkid * epb + i);
|
2013-11-13 19:05:17 +00:00
|
|
|
traverse_prefetch_metadata(td,
|
|
|
|
&((blkptr_t *)buf->b_data)[i], czb);
|
2013-07-02 20:20:02 +00:00
|
|
|
}
|
2008-12-03 20:09:06 +00:00
|
|
|
|
|
|
|
/* recursively visitbp() blocks below this */
|
2013-07-02 20:20:02 +00:00
|
|
|
for (i = 0; i < epb; i++) {
|
2013-11-13 19:05:17 +00:00
|
|
|
SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object,
|
2011-05-25 23:09:57 +00:00
|
|
|
zb->zb_level - 1,
|
|
|
|
zb->zb_blkid * epb + i);
|
2013-11-13 19:05:17 +00:00
|
|
|
err = traverse_visitbp(td, dnp,
|
|
|
|
&((blkptr_t *)buf->b_data)[i], czb);
|
2014-06-05 21:20:08 +00:00
|
|
|
if (err != 0)
|
|
|
|
break;
|
2008-12-03 20:09:06 +00:00
|
|
|
}
|
2013-11-13 19:05:17 +00:00
|
|
|
|
2014-06-25 18:37:59 +00:00
|
|
|
kmem_free(czb, sizeof (zbookmark_phys_t));
|
2013-11-13 19:05:17 +00:00
|
|
|
|
2011-05-25 23:09:57 +00:00
|
|
|
} else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) {
|
|
|
|
uint32_t flags = ARC_WAIT;
|
2013-11-13 19:05:17 +00:00
|
|
|
int32_t i;
|
|
|
|
int32_t epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT;
|
2015-04-26 08:09:41 +00:00
|
|
|
dnode_phys_t *cdnp;
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
|
2011-05-25 23:09:57 +00:00
|
|
|
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0)
|
2014-06-05 21:20:08 +00:00
|
|
|
goto post;
|
2015-04-26 08:09:41 +00:00
|
|
|
cdnp = buf->b_data;
|
2013-07-02 20:20:02 +00:00
|
|
|
|
|
|
|
for (i = 0; i < epb; i++) {
|
2015-04-26 08:09:41 +00:00
|
|
|
prefetch_dnode_metadata(td, &cdnp[i], zb->zb_objset,
|
2013-07-02 20:20:02 +00:00
|
|
|
zb->zb_blkid * epb + i);
|
|
|
|
}
|
2008-12-03 20:09:06 +00:00
|
|
|
|
|
|
|
/* recursively visitbp() blocks below this */
|
2013-07-02 20:20:02 +00:00
|
|
|
for (i = 0; i < epb; i++) {
|
2015-04-26 08:09:41 +00:00
|
|
|
err = traverse_dnode(td, &cdnp[i], zb->zb_objset,
|
2011-05-25 23:09:57 +00:00
|
|
|
zb->zb_blkid * epb + i);
|
2014-06-05 21:20:08 +00:00
|
|
|
if (err != 0)
|
|
|
|
break;
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
2011-05-25 23:09:57 +00:00
|
|
|
} else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) {
|
|
|
|
uint32_t flags = ARC_WAIT;
|
|
|
|
objset_phys_t *osp;
|
2015-04-26 08:09:41 +00:00
|
|
|
dnode_phys_t *mdnp, *gdnp, *udnp;
|
2011-05-25 23:09:57 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
|
2011-05-25 23:09:57 +00:00
|
|
|
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0)
|
2014-06-05 21:20:08 +00:00
|
|
|
goto post;
|
2011-05-25 23:09:57 +00:00
|
|
|
|
|
|
|
osp = buf->b_data;
|
2015-04-26 08:09:41 +00:00
|
|
|
mdnp = &osp->os_meta_dnode;
|
|
|
|
gdnp = &osp->os_groupused_dnode;
|
|
|
|
udnp = &osp->os_userused_dnode;
|
|
|
|
|
|
|
|
prefetch_dnode_metadata(td, mdnp, zb->zb_objset,
|
2013-07-02 20:20:02 +00:00
|
|
|
DMU_META_DNODE_OBJECT);
|
|
|
|
if (arc_buf_size(buf) >= sizeof (objset_phys_t)) {
|
2015-04-26 08:09:41 +00:00
|
|
|
prefetch_dnode_metadata(td, gdnp, zb->zb_objset,
|
|
|
|
DMU_GROUPUSED_OBJECT);
|
|
|
|
prefetch_dnode_metadata(td, udnp, zb->zb_objset,
|
|
|
|
DMU_USERUSED_OBJECT);
|
2013-07-02 20:20:02 +00:00
|
|
|
}
|
|
|
|
|
2015-04-26 08:09:41 +00:00
|
|
|
err = traverse_dnode(td, mdnp, zb->zb_objset,
|
2011-05-25 23:09:57 +00:00
|
|
|
DMU_META_DNODE_OBJECT);
|
|
|
|
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
|
2015-04-26 08:09:41 +00:00
|
|
|
err = traverse_dnode(td, gdnp, zb->zb_objset,
|
2014-01-24 15:54:37 +00:00
|
|
|
DMU_GROUPUSED_OBJECT);
|
2009-07-02 22:44:48 +00:00
|
|
|
}
|
2011-05-25 23:09:57 +00:00
|
|
|
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
|
2015-04-26 08:09:41 +00:00
|
|
|
err = traverse_dnode(td, udnp, zb->zb_objset,
|
2014-01-24 15:54:37 +00:00
|
|
|
DMU_USERUSED_OBJECT);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-05-25 23:09:57 +00:00
|
|
|
if (buf)
|
|
|
|
(void) arc_buf_remove_ref(buf, &buf);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2012-12-13 23:24:15 +00:00
|
|
|
post:
|
2014-06-05 21:20:08 +00:00
|
|
|
if (err == 0 && (td->td_flags & TRAVERSE_POST))
|
2013-07-02 20:26:24 +00:00
|
|
|
err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg);
|
2014-06-05 21:20:08 +00:00
|
|
|
|
|
|
|
if ((td->td_flags & TRAVERSE_HARD) && (err == EIO || err == ECKSUM)) {
|
|
|
|
/*
|
|
|
|
* Ignore this disk error as requested by the HARD flag,
|
|
|
|
* and continue traversal.
|
|
|
|
*/
|
|
|
|
err = 0;
|
2012-12-13 23:24:15 +00:00
|
|
|
}
|
|
|
|
|
2014-06-05 21:20:08 +00:00
|
|
|
/*
|
|
|
|
* If we are stopping here, set td_resume.
|
|
|
|
*/
|
|
|
|
if (td->td_resume != NULL && err != 0 && !td->td_paused) {
|
|
|
|
td->td_resume->zb_objset = zb->zb_objset;
|
|
|
|
td->td_resume->zb_object = zb->zb_object;
|
|
|
|
td->td_resume->zb_level = 0;
|
|
|
|
/*
|
|
|
|
* If we have stopped on an indirect block (e.g. due to
|
|
|
|
* i/o error), we have not visited anything below it.
|
|
|
|
* Set the bookmark to the first level-0 block that we need
|
|
|
|
* to visit. This way, the resuming code does not need to
|
|
|
|
* deal with resuming from indirect blocks.
|
|
|
|
*/
|
|
|
|
td->td_resume->zb_blkid = zb->zb_blkid <<
|
|
|
|
(zb->zb_level * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT));
|
|
|
|
td->td_paused = B_TRUE;
|
2010-05-28 20:45:14 +00:00
|
|
|
}
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2014-06-05 21:20:08 +00:00
|
|
|
return (err);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2013-07-02 20:20:02 +00:00
|
|
|
static void
|
|
|
|
prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *dnp,
|
2013-07-02 20:26:24 +00:00
|
|
|
uint64_t objset, uint64_t object)
|
2013-07-02 20:20:02 +00:00
|
|
|
{
|
|
|
|
int j;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t czb;
|
2013-07-02 20:20:02 +00:00
|
|
|
|
|
|
|
for (j = 0; j < dnp->dn_nblkptr; j++) {
|
|
|
|
SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j);
|
2013-07-02 20:26:24 +00:00
|
|
|
traverse_prefetch_metadata(td, &dnp->dn_blkptr[j], &czb);
|
2013-07-02 20:20:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
|
|
|
|
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
|
2013-07-02 20:26:24 +00:00
|
|
|
traverse_prefetch_metadata(td, &dnp->dn_spill, &czb);
|
2013-07-02 20:20:02 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-07-02 22:44:48 +00:00
|
|
|
static int
|
2010-08-26 21:24:34 +00:00
|
|
|
traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
|
2013-07-02 20:26:24 +00:00
|
|
|
uint64_t objset, uint64_t object)
|
2009-07-02 22:44:48 +00:00
|
|
|
{
|
2014-06-05 21:20:08 +00:00
|
|
|
int j, err = 0;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t czb;
|
2009-07-02 22:44:48 +00:00
|
|
|
|
|
|
|
for (j = 0; j < dnp->dn_nblkptr; j++) {
|
|
|
|
SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j);
|
2013-07-02 20:26:24 +00:00
|
|
|
err = traverse_visitbp(td, dnp, &dnp->dn_blkptr[j], &czb);
|
2014-06-05 21:20:08 +00:00
|
|
|
if (err != 0)
|
|
|
|
break;
|
2009-07-02 22:44:48 +00:00
|
|
|
}
|
2010-05-28 20:45:14 +00:00
|
|
|
|
2014-11-23 03:14:24 +00:00
|
|
|
if (err == 0 && dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
|
2013-07-02 20:20:02 +00:00
|
|
|
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
|
2013-07-02 20:26:24 +00:00
|
|
|
err = traverse_visitbp(td, dnp, &dnp->dn_spill, &czb);
|
2010-05-28 20:45:14 +00:00
|
|
|
}
|
2014-06-05 21:20:08 +00:00
|
|
|
return (err);
|
2009-07-02 22:44:48 +00:00
|
|
|
}
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
/* ARGSUSED */
|
|
|
|
static int
|
2010-05-28 20:45:14 +00:00
|
|
|
traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
|
2014-06-25 18:37:59 +00:00
|
|
|
const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
2010-08-26 21:24:34 +00:00
|
|
|
prefetch_data_t *pfd = arg;
|
2008-12-03 20:09:06 +00:00
|
|
|
uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2015-03-27 04:31:52 +00:00
|
|
|
ASSERT(pfd->pd_bytes_fetched >= 0);
|
2008-12-03 20:09:06 +00:00
|
|
|
if (pfd->pd_cancel)
|
2013-03-08 18:41:28 +00:00
|
|
|
return (SET_ERROR(EINTR));
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2014-06-05 21:34:21 +00:00
|
|
|
if (!prefetch_needed(pfd, bp))
|
2008-11-20 20:01:55 +00:00
|
|
|
return (0);
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
mutex_enter(&pfd->pd_mtx);
|
2015-03-27 04:31:52 +00:00
|
|
|
while (!pfd->pd_cancel && pfd->pd_bytes_fetched >= zfs_pd_bytes_max)
|
2008-12-03 20:09:06 +00:00
|
|
|
cv_wait(&pfd->pd_cv, &pfd->pd_mtx);
|
2015-03-27 04:31:52 +00:00
|
|
|
pfd->pd_bytes_fetched += BP_GET_LSIZE(bp);
|
2008-12-03 20:09:06 +00:00
|
|
|
cv_broadcast(&pfd->pd_cv);
|
|
|
|
mutex_exit(&pfd->pd_mtx);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
(void) arc_read(NULL, spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ,
|
|
|
|
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, zb);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
return (0);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2008-12-03 20:09:06 +00:00
|
|
|
traverse_prefetch_thread(void *arg)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
2010-08-26 21:24:34 +00:00
|
|
|
traverse_data_t *td_main = arg;
|
|
|
|
traverse_data_t td = *td_main;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t czb;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
td.td_func = traverse_prefetcher;
|
|
|
|
td.td_arg = td_main->td_pfd;
|
|
|
|
td.td_pfd = NULL;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2010-05-28 20:45:14 +00:00
|
|
|
SET_BOOKMARK(&czb, td.td_objset,
|
|
|
|
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
|
2013-07-02 20:26:24 +00:00
|
|
|
(void) traverse_visitbp(&td, NULL, td.td_rootbp, &czb);
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
mutex_enter(&td_main->td_pfd->pd_mtx);
|
|
|
|
td_main->td_pfd->pd_exited = B_TRUE;
|
|
|
|
cv_broadcast(&td_main->td_pfd->pd_cv);
|
|
|
|
mutex_exit(&td_main->td_pfd->pd_mtx);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
/*
|
|
|
|
* NB: dataset must not be changing on-disk (eg, is a snapshot or we are
|
|
|
|
* in syncing context).
|
|
|
|
*/
|
|
|
|
static int
|
2012-12-13 23:24:15 +00:00
|
|
|
traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp,
|
2014-06-25 18:37:59 +00:00
|
|
|
uint64_t txg_start, zbookmark_phys_t *resume, int flags,
|
2012-12-13 23:24:15 +00:00
|
|
|
blkptr_cb_t func, void *arg)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
2010-08-26 17:58:47 +00:00
|
|
|
traverse_data_t *td;
|
|
|
|
prefetch_data_t *pd;
|
2014-06-25 18:37:59 +00:00
|
|
|
zbookmark_phys_t *czb;
|
2008-12-03 20:09:06 +00:00
|
|
|
int err;
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2012-12-13 23:24:15 +00:00
|
|
|
ASSERT(ds == NULL || objset == ds->ds_object);
|
|
|
|
ASSERT(!(flags & TRAVERSE_PRE) || !(flags & TRAVERSE_POST));
|
|
|
|
|
2013-07-02 20:20:02 +00:00
|
|
|
/*
|
|
|
|
* The data prefetching mechanism (the prefetch thread) is incompatible
|
|
|
|
* with resuming from a bookmark.
|
|
|
|
*/
|
|
|
|
ASSERT(resume == NULL || !(flags & TRAVERSE_PREFETCH_DATA));
|
|
|
|
|
2014-11-21 00:09:39 +00:00
|
|
|
td = kmem_alloc(sizeof (traverse_data_t), KM_SLEEP);
|
|
|
|
pd = kmem_zalloc(sizeof (prefetch_data_t), KM_SLEEP);
|
|
|
|
czb = kmem_alloc(sizeof (zbookmark_phys_t), KM_SLEEP);
|
2010-08-26 17:58:47 +00:00
|
|
|
|
|
|
|
td->td_spa = spa;
|
2012-12-13 23:24:15 +00:00
|
|
|
td->td_objset = objset;
|
2010-08-26 17:58:47 +00:00
|
|
|
td->td_rootbp = rootbp;
|
|
|
|
td->td_min_txg = txg_start;
|
2012-12-13 23:24:15 +00:00
|
|
|
td->td_resume = resume;
|
2010-08-26 17:58:47 +00:00
|
|
|
td->td_func = func;
|
|
|
|
td->td_arg = arg;
|
|
|
|
td->td_pfd = pd;
|
|
|
|
td->td_flags = flags;
|
2014-06-05 21:20:08 +00:00
|
|
|
td->td_paused = B_FALSE;
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2010-08-26 17:58:47 +00:00
|
|
|
pd->pd_flags = flags;
|
|
|
|
mutex_init(&pd->pd_mtx, NULL, MUTEX_DEFAULT, NULL);
|
|
|
|
cv_init(&pd->pd_cv, NULL, CV_DEFAULT, NULL);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
Add visibility in to arc_read
This change is an attempt to add visibility into the arc_read calls
occurring on a system, in real time. To do this, a list was added to the
in memory SPA data structure for a pool, with each element on the list
corresponding to a call to arc_read. These entries are then exported
through the kstat interface, which can then be interpreted in userspace.
For each arc_read call, the following information is exported:
* A unique identifier (uint64_t)
* The time the entry was added to the list (hrtime_t)
(*not* wall clock time; relative to the other entries on the list)
* The objset ID (uint64_t)
* The object number (uint64_t)
* The indirection level (uint64_t)
* The block ID (uint64_t)
* The name of the function originating the arc_read call (char[24])
* The arc_flags from the arc_read call (uint32_t)
* The PID of the reading thread (pid_t)
* The command or name of thread originating read (char[16])
From this exported information one can see, in real time, exactly what
is being read, what function is generating the read, and whether or not
the read was found to be already cached.
There is still some work to be done, but this should serve as a good
starting point.
Specifically, dbuf_read's are not accounted for in the currently
exported information. Thus, a follow up patch should probably be added
to export these calls that never call into arc_read (they only hit the
dbuf hash table). In addition, it might be nice to create a utility
similar to "arcstat.py" to digest the exported information and display
it in a more readable format. Or perhaps, log the information and allow
for it to be "replayed" at a later time.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2013-09-06 23:09:05 +00:00
|
|
|
SET_BOOKMARK(czb, td->td_objset,
|
|
|
|
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
|
|
|
|
|
2010-08-26 21:24:34 +00:00
|
|
|
/* See comment on ZIL traversal in dsl_scan_visitds. */
|
2013-09-04 12:00:57 +00:00
|
|
|
if (ds != NULL && !dsl_dataset_is_snapshot(ds) && !BP_IS_HOLE(rootbp)) {
|
|
|
|
uint32_t flags = ARC_WAIT;
|
|
|
|
objset_phys_t *osp;
|
|
|
|
arc_buf_t *buf;
|
2010-08-26 21:24:34 +00:00
|
|
|
|
2013-09-04 12:00:57 +00:00
|
|
|
err = arc_read(NULL, td->td_spa, rootbp,
|
|
|
|
arc_getbuf_func, &buf,
|
Add visibility in to arc_read
This change is an attempt to add visibility into the arc_read calls
occurring on a system, in real time. To do this, a list was added to the
in memory SPA data structure for a pool, with each element on the list
corresponding to a call to arc_read. These entries are then exported
through the kstat interface, which can then be interpreted in userspace.
For each arc_read call, the following information is exported:
* A unique identifier (uint64_t)
* The time the entry was added to the list (hrtime_t)
(*not* wall clock time; relative to the other entries on the list)
* The objset ID (uint64_t)
* The object number (uint64_t)
* The indirection level (uint64_t)
* The block ID (uint64_t)
* The name of the function originating the arc_read call (char[24])
* The arc_flags from the arc_read call (uint32_t)
* The PID of the reading thread (pid_t)
* The command or name of thread originating read (char[16])
From this exported information one can see, in real time, exactly what
is being read, what function is generating the read, and whether or not
the read was found to be already cached.
There is still some work to be done, but this should serve as a good
starting point.
Specifically, dbuf_read's are not accounted for in the currently
exported information. Thus, a follow up patch should probably be added
to export these calls that never call into arc_read (they only hit the
dbuf hash table). In addition, it might be nice to create a utility
similar to "arcstat.py" to digest the exported information and display
it in a more readable format. Or perhaps, log the information and allow
for it to be "replayed" at a later time.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2013-09-06 23:09:05 +00:00
|
|
|
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, czb);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0)
|
2010-08-26 21:24:34 +00:00
|
|
|
return (err);
|
|
|
|
|
2013-09-04 12:00:57 +00:00
|
|
|
osp = buf->b_data;
|
|
|
|
traverse_zil(td, &osp->os_zil_header);
|
|
|
|
(void) arc_buf_remove_ref(buf, &buf);
|
2010-08-26 21:24:34 +00:00
|
|
|
}
|
|
|
|
|
2013-07-02 20:20:02 +00:00
|
|
|
if (!(flags & TRAVERSE_PREFETCH_DATA) ||
|
2008-12-03 20:09:06 +00:00
|
|
|
0 == taskq_dispatch(system_taskq, traverse_prefetch_thread,
|
2010-08-26 17:58:47 +00:00
|
|
|
td, TQ_NOQUEUE))
|
|
|
|
pd->pd_exited = B_TRUE;
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2013-07-02 20:26:24 +00:00
|
|
|
err = traverse_visitbp(td, NULL, rootbp, czb);
|
2010-08-26 17:58:47 +00:00
|
|
|
|
|
|
|
mutex_enter(&pd->pd_mtx);
|
|
|
|
pd->pd_cancel = B_TRUE;
|
|
|
|
cv_broadcast(&pd->pd_cv);
|
|
|
|
while (!pd->pd_exited)
|
|
|
|
cv_wait(&pd->pd_cv, &pd->pd_mtx);
|
|
|
|
mutex_exit(&pd->pd_mtx);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2010-08-26 17:58:47 +00:00
|
|
|
mutex_destroy(&pd->pd_mtx);
|
|
|
|
cv_destroy(&pd->pd_cv);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2014-06-25 18:37:59 +00:00
|
|
|
kmem_free(czb, sizeof (zbookmark_phys_t));
|
2013-11-13 19:05:17 +00:00
|
|
|
kmem_free(pd, sizeof (struct prefetch_data));
|
|
|
|
kmem_free(td, sizeof (struct traverse_data));
|
2008-11-20 20:01:55 +00:00
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
return (err);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
/*
|
|
|
|
* NB: dataset must not be changing on-disk (eg, is a snapshot or we are
|
|
|
|
* in syncing context).
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
traverse_dataset(dsl_dataset_t *ds, uint64_t txg_start, int flags,
|
|
|
|
blkptr_cb_t func, void *arg)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
2012-12-13 23:24:15 +00:00
|
|
|
return (traverse_impl(ds->ds_dir->dd_pool->dp_spa, ds, ds->ds_object,
|
|
|
|
&ds->ds_phys->ds_bp, txg_start, NULL, flags, func, arg));
|
|
|
|
}
|
|
|
|
|
|
|
|
int
|
|
|
|
traverse_dataset_destroyed(spa_t *spa, blkptr_t *blkptr,
|
2014-06-25 18:37:59 +00:00
|
|
|
uint64_t txg_start, zbookmark_phys_t *resume, int flags,
|
2012-12-13 23:24:15 +00:00
|
|
|
blkptr_cb_t func, void *arg)
|
|
|
|
{
|
|
|
|
return (traverse_impl(spa, NULL, ZB_DESTROYED_OBJSET,
|
|
|
|
blkptr, txg_start, resume, flags, func, arg));
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
|
|
|
|
2008-12-03 20:09:06 +00:00
|
|
|
/*
|
|
|
|
* NB: pool must not be changing on-disk (eg, from zdb or sync context).
|
|
|
|
*/
|
|
|
|
int
|
2010-05-28 20:45:14 +00:00
|
|
|
traverse_pool(spa_t *spa, uint64_t txg_start, int flags,
|
|
|
|
blkptr_cb_t func, void *arg)
|
2008-11-20 20:01:55 +00:00
|
|
|
{
|
2014-06-05 21:20:08 +00:00
|
|
|
int err;
|
2008-12-03 20:09:06 +00:00
|
|
|
uint64_t obj;
|
|
|
|
dsl_pool_t *dp = spa_get_dsl(spa);
|
|
|
|
objset_t *mos = dp->dp_meta_objset;
|
2010-05-28 20:45:14 +00:00
|
|
|
boolean_t hard = (flags & TRAVERSE_HARD);
|
2008-12-03 20:09:06 +00:00
|
|
|
|
|
|
|
/* visit the MOS */
|
2012-12-13 23:24:15 +00:00
|
|
|
err = traverse_impl(spa, NULL, 0, spa_get_rootblkptr(spa),
|
|
|
|
txg_start, NULL, flags, func, arg);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0)
|
2008-12-03 20:09:06 +00:00
|
|
|
return (err);
|
|
|
|
|
|
|
|
/* visit each dataset */
|
2014-06-05 21:20:08 +00:00
|
|
|
for (obj = 1; err == 0;
|
2010-05-28 20:45:14 +00:00
|
|
|
err = dmu_object_next(mos, &obj, FALSE, txg_start)) {
|
2008-12-03 20:09:06 +00:00
|
|
|
dmu_object_info_t doi;
|
|
|
|
|
|
|
|
err = dmu_object_info(mos, obj, &doi);
|
2013-09-04 12:00:57 +00:00
|
|
|
if (err != 0) {
|
2014-06-05 21:20:08 +00:00
|
|
|
if (hard)
|
|
|
|
continue;
|
|
|
|
break;
|
2010-05-28 20:45:14 +00:00
|
|
|
}
|
2008-12-03 20:09:06 +00:00
|
|
|
|
2013-10-08 17:13:05 +00:00
|
|
|
if (doi.doi_bonus_type == DMU_OT_DSL_DATASET) {
|
2008-12-03 20:09:06 +00:00
|
|
|
dsl_dataset_t *ds;
|
2010-05-28 20:45:14 +00:00
|
|
|
uint64_t txg = txg_start;
|
|
|
|
|
2013-09-04 12:00:57 +00:00
|
|
|
dsl_pool_config_enter(dp, FTAG);
|
2008-12-03 20:09:06 +00:00
|
|
|
err = dsl_dataset_hold_obj(dp, obj, FTAG, &ds);
|
2013-09-04 12:00:57 +00:00
|
|
|
dsl_pool_config_exit(dp, FTAG);
|
|
|
|
if (err != 0) {
|
2014-06-05 21:20:08 +00:00
|
|
|
if (hard)
|
|
|
|
continue;
|
|
|
|
break;
|
2010-05-28 20:45:14 +00:00
|
|
|
}
|
|
|
|
if (ds->ds_phys->ds_prev_snap_txg > txg)
|
|
|
|
txg = ds->ds_phys->ds_prev_snap_txg;
|
|
|
|
err = traverse_dataset(ds, txg, flags, func, arg);
|
2008-12-03 20:09:06 +00:00
|
|
|
dsl_dataset_rele(ds, FTAG);
|
2014-06-05 21:20:08 +00:00
|
|
|
if (err != 0)
|
|
|
|
break;
|
2008-12-03 20:09:06 +00:00
|
|
|
}
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
2008-12-03 20:09:06 +00:00
|
|
|
if (err == ESRCH)
|
|
|
|
err = 0;
|
2014-06-05 21:20:08 +00:00
|
|
|
return (err);
|
2008-11-20 20:01:55 +00:00
|
|
|
}
|
2010-08-26 18:49:16 +00:00
|
|
|
|
|
|
|
#if defined(_KERNEL) && defined(HAVE_SPL)
|
|
|
|
EXPORT_SYMBOL(traverse_dataset);
|
|
|
|
EXPORT_SYMBOL(traverse_pool);
|
2011-05-03 22:09:28 +00:00
|
|
|
|
2015-03-27 04:31:52 +00:00
|
|
|
module_param(zfs_pd_bytes_max, int, 0644);
|
|
|
|
MODULE_PARM_DESC(zfs_pd_bytes_max, "Max number of bytes to prefetch");
|
2010-08-26 18:49:16 +00:00
|
|
|
#endif
|