freebsd-dev/module/zfs/dmu_traverse.c

668 lines
18 KiB
C
Raw Normal View History

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
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* 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>
#include <sys/sa.h>
#include <sys/sa_impl.h>
#include <sys/callb.h>
#include <sys/zfeature.h>
int zfs_pd_blks_max = 100;
typedef struct prefetch_data {
kmutex_t pd_mtx;
kcondvar_t pd_cv;
int pd_blks_max;
int pd_blks_fetched;
int pd_flags;
boolean_t pd_cancel;
boolean_t pd_exited;
} prefetch_data_t;
typedef struct traverse_data {
spa_t *td_spa;
uint64_t td_objset;
blkptr_t *td_rootbp;
uint64_t td_min_txg;
zbookmark_phys_t *td_resume;
int td_flags;
prefetch_data_t *td_pfd;
boolean_t td_paused;
blkptr_cb_t *td_func;
void *td_arg;
} traverse_data_t;
2008-11-20 20:01:55 +00:00
static int traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object);
static void prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *,
uint64_t objset, uint64_t object);
2009-07-02 22:44:48 +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)
{
traverse_data_t *td = arg;
zbookmark_phys_t zb;
2008-11-20 20:01:55 +00:00
if (BP_IS_HOLE(bp))
return (0);
2008-11-20 20:01:55 +00:00
if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(td->td_spa))
return (0);
SET_BOOKMARK(&zb, td->td_objset, ZB_ZIL_OBJECT, ZB_ZIL_LEVEL,
bp->blk_cksum.zc_word[ZIL_ZC_SEQ]);
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL, td->td_arg);
return (0);
2008-11-20 20:01:55 +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)
{
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;
zbookmark_phys_t zb;
2008-11-20 20:01:55 +00:00
if (BP_IS_HOLE(bp))
return (0);
2008-11-20 20:01:55 +00:00
if (claim_txg == 0 || bp->blk_birth < claim_txg)
return (0);
SET_BOOKMARK(&zb, td->td_objset, lr->lr_foid,
ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp));
(void) td->td_func(td->td_spa, zilog, bp, &zb, NULL,
td->td_arg);
2008-11-20 20:01:55 +00:00
}
return (0);
2008-11-20 20:01:55 +00:00
}
static void
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
* 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;
zilog = zil_alloc(spa_get_dsl(td->td_spa)->dp_meta_objset, zh);
2008-11-20 20:01:55 +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);
}
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,
const zbookmark_phys_t *zb)
{
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);
}
static void
traverse_prefetch_metadata(traverse_data_t *td,
const blkptr_t *bp, const zbookmark_phys_t *zb)
{
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;
(void) arc_read(NULL, td->td_spa, bp, NULL, NULL,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
}
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);
}
static int
traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
const blkptr_t *bp, const zbookmark_phys_t *zb)
{
int err = 0;
arc_buf_t *buf = NULL;
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);
}
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);
}
if (td->td_pfd != NULL && !td->td_pfd->pd_exited &&
prefetch_needed(td->td_pfd, bp)) {
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
mutex_enter(&td->td_pfd->pd_mtx);
ASSERT(td->td_pfd->pd_blks_fetched >= 0);
while (td->td_pfd->pd_blks_fetched == 0 &&
!td->td_pfd->pd_exited)
cv_wait(&td->td_pfd->pd_cv, &td->td_pfd->pd_mtx);
td->td_pfd->pd_blks_fetched--;
cv_broadcast(&td->td_pfd->pd_cv);
mutex_exit(&td->td_pfd->pd_mtx);
2008-11-20 20:01:55 +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);
}
if (td->td_flags & TRAVERSE_PRE) {
err = td->td_func(td->td_spa, NULL, bp, zb, dnp,
td->td_arg);
if (err == TRAVERSE_VISIT_NO_CHILDREN)
return (0);
if (err != 0)
goto post;
2008-11-20 20:01:55 +00:00
}
if (BP_GET_LEVEL(bp) > 0) {
uint32_t flags = ARC_WAIT;
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
int32_t i;
int32_t epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT;
zbookmark_phys_t *czb;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err != 0)
goto post;
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
czb = kmem_alloc(sizeof (zbookmark_phys_t), KM_PUSHPAGE);
for (i = 0; i < epb; i++) {
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object,
zb->zb_level - 1,
zb->zb_blkid * epb + i);
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
traverse_prefetch_metadata(td,
&((blkptr_t *)buf->b_data)[i], czb);
}
/* recursively visitbp() blocks below this */
for (i = 0; i < epb; i++) {
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
SET_BOOKMARK(czb, zb->zb_objset, zb->zb_object,
zb->zb_level - 1,
zb->zb_blkid * epb + i);
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
err = traverse_visitbp(td, dnp,
&((blkptr_t *)buf->b_data)[i], czb);
if (err != 0)
break;
}
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
kmem_free(czb, sizeof (zbookmark_phys_t));
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
} else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) {
uint32_t flags = ARC_WAIT;
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
int32_t i;
int32_t epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err != 0)
goto post;
dnp = buf->b_data;
for (i = 0; i < epb; i++) {
prefetch_dnode_metadata(td, &dnp[i], zb->zb_objset,
zb->zb_blkid * epb + i);
}
/* recursively visitbp() blocks below this */
for (i = 0; i < epb; i++) {
err = traverse_dnode(td, &dnp[i], zb->zb_objset,
zb->zb_blkid * epb + i);
if (err != 0)
break;
2008-11-20 20:01:55 +00:00
}
} else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) {
uint32_t flags = ARC_WAIT;
objset_phys_t *osp;
dnode_phys_t *dnp;
err = arc_read(NULL, td->td_spa, bp, arc_getbuf_func, &buf,
ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
if (err != 0)
goto post;
osp = buf->b_data;
dnp = &osp->os_meta_dnode;
prefetch_dnode_metadata(td, dnp, zb->zb_objset,
DMU_META_DNODE_OBJECT);
if (arc_buf_size(buf) >= sizeof (objset_phys_t)) {
prefetch_dnode_metadata(td, &osp->os_groupused_dnode,
zb->zb_objset, DMU_GROUPUSED_OBJECT);
prefetch_dnode_metadata(td, &osp->os_userused_dnode,
zb->zb_objset, DMU_USERUSED_OBJECT);
}
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_META_DNODE_OBJECT);
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
dnp = &osp->os_groupused_dnode;
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_GROUPUSED_OBJECT);
2009-07-02 22:44:48 +00:00
}
if (err == 0 && arc_buf_size(buf) >= sizeof (objset_phys_t)) {
dnp = &osp->os_userused_dnode;
err = traverse_dnode(td, dnp, zb->zb_objset,
DMU_USERUSED_OBJECT);
2008-11-20 20:01:55 +00:00
}
}
if (buf)
(void) arc_buf_remove_ref(buf, &buf);
2008-11-20 20:01:55 +00:00
post:
if (err == 0 && (td->td_flags & TRAVERSE_POST))
err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg);
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;
}
/*
* 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;
}
2008-11-20 20:01:55 +00:00
return (err);
2008-11-20 20:01:55 +00:00
}
static void
prefetch_dnode_metadata(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object)
{
int j;
zbookmark_phys_t czb;
for (j = 0; j < dnp->dn_nblkptr; j++) {
SET_BOOKMARK(&czb, objset, object, dnp->dn_nlevels - 1, j);
traverse_prefetch_metadata(td, &dnp->dn_blkptr[j], &czb);
}
if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
traverse_prefetch_metadata(td, &dnp->dn_spill, &czb);
}
}
2009-07-02 22:44:48 +00:00
static int
traverse_dnode(traverse_data_t *td, const dnode_phys_t *dnp,
uint64_t objset, uint64_t object)
2009-07-02 22:44:48 +00:00
{
int j, err = 0;
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);
err = traverse_visitbp(td, dnp, &dnp->dn_blkptr[j], &czb);
if (err != 0)
break;
2009-07-02 22:44:48 +00:00
}
if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
SET_BOOKMARK(&czb, objset, object, 0, DMU_SPILL_BLKID);
err = traverse_visitbp(td, dnp, &dnp->dn_spill, &czb);
}
return (err);
2009-07-02 22:44:48 +00:00
}
/* ARGSUSED */
static int
traverse_prefetcher(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
2008-11-20 20:01:55 +00:00
{
prefetch_data_t *pfd = arg;
uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH;
2008-11-20 20:01:55 +00:00
ASSERT(pfd->pd_blks_fetched >= 0);
if (pfd->pd_cancel)
return (SET_ERROR(EINTR));
2008-11-20 20:01:55 +00:00
if (!prefetch_needed(pfd, bp))
2008-11-20 20:01:55 +00:00
return (0);
mutex_enter(&pfd->pd_mtx);
while (!pfd->pd_cancel && pfd->pd_blks_fetched >= pfd->pd_blks_max)
cv_wait(&pfd->pd_cv, &pfd->pd_mtx);
pfd->pd_blks_fetched++;
cv_broadcast(&pfd->pd_cv);
mutex_exit(&pfd->pd_mtx);
2008-11-20 20:01:55 +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
return (0);
2008-11-20 20:01:55 +00:00
}
static void
traverse_prefetch_thread(void *arg)
2008-11-20 20:01:55 +00:00
{
traverse_data_t *td_main = arg;
traverse_data_t td = *td_main;
zbookmark_phys_t czb;
2008-11-20 20:01:55 +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
SET_BOOKMARK(&czb, td.td_objset,
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
(void) traverse_visitbp(&td, NULL, td.td_rootbp, &czb);
2008-11-20 20:01:55 +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
}
/*
* NB: dataset must not be changing on-disk (eg, is a snapshot or we are
* in syncing context).
*/
static int
traverse_impl(spa_t *spa, dsl_dataset_t *ds, uint64_t objset, blkptr_t *rootbp,
uint64_t txg_start, zbookmark_phys_t *resume, int flags,
blkptr_cb_t func, void *arg)
2008-11-20 20:01:55 +00:00
{
traverse_data_t *td;
prefetch_data_t *pd;
zbookmark_phys_t *czb;
int err;
2008-11-20 20:01:55 +00:00
ASSERT(ds == NULL || objset == ds->ds_object);
ASSERT(!(flags & TRAVERSE_PRE) || !(flags & TRAVERSE_POST));
/*
* The data prefetching mechanism (the prefetch thread) is incompatible
* with resuming from a bookmark.
*/
ASSERT(resume == NULL || !(flags & TRAVERSE_PREFETCH_DATA));
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
2013-11-13 19:05:17 +00:00
td = kmem_alloc(sizeof (traverse_data_t), KM_PUSHPAGE);
pd = kmem_zalloc(sizeof (prefetch_data_t), KM_PUSHPAGE);
czb = kmem_alloc(sizeof (zbookmark_phys_t), KM_PUSHPAGE);
td->td_spa = spa;
td->td_objset = objset;
td->td_rootbp = rootbp;
td->td_min_txg = txg_start;
td->td_resume = resume;
td->td_func = func;
td->td_arg = arg;
td->td_pfd = pd;
td->td_flags = flags;
td->td_paused = B_FALSE;
pd->pd_blks_max = zfs_pd_blks_max;
pd->pd_flags = flags;
mutex_init(&pd->pd_mtx, NULL, MUTEX_DEFAULT, NULL);
cv_init(&pd->pd_cv, NULL, CV_DEFAULT, NULL);
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);
/* See comment on ZIL traversal in dsl_scan_visitds. */
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;
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);
if (err != 0)
return (err);
osp = buf->b_data;
traverse_zil(td, &osp->os_zil_header);
(void) arc_buf_remove_ref(buf, &buf);
}
if (!(flags & TRAVERSE_PREFETCH_DATA) ||
0 == taskq_dispatch(system_taskq, traverse_prefetch_thread,
td, TQ_NOQUEUE))
pd->pd_exited = B_TRUE;
err = traverse_visitbp(td, NULL, rootbp, czb);
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);
mutex_destroy(&pd->pd_mtx);
cv_destroy(&pd->pd_cv);
kmem_free(czb, sizeof (zbookmark_phys_t));
Reduce stack for traverse_visitbp() recursion During pool import stack overflows may still occur due to the potentially deep recursion of traverse_visitbp(). This is most likely to occur when additional layers are added to the block device stack such as DM multipath. To minimize the stack usage for this call path the following changes were made: 1) Added the keywork 'noinline' to the vdev_*_map_alloc() functions to prevent them from being inlined by gcc. This reduced the stack usage of vdev_raidz_io_start() from 208 to 128 bytes, and vdev_mirror_io_start() from 144 to 128 bytes. 2) The 'saved_poolname' charater array in zfsdev_ioctl() was moved from the stack to the heap. This reduced the stack usage of zfsdev_ioctl() from 368 to 112 bytes. 3) The major saving came from slimming down traverse_visitbp() from from 224 to 144 bytes. Since this function is called recursively the 80 bytes saved per invokation adds up. The following changes were made: a) The 'hard' local variable was replaced by a TD_HARD() macro. b) The 'pd' local variable was replaced by 'td->td_pfd' references. c) The zbookmark_t was moved to the heap. This does cost us an additional memory allocation per recursion by that cost should still be minimal. The cost could be further reduced by adding a dedicated zbookmark_t slab cache. d) The variable declarations in 'if (BP_GET_LEVEL()) { }' were restructured to use the minimum amount of stack. This includes removing the 'cbp' local variable. Overall for the offending use case roughly 1584 of total stack space has been saved. This is enough to avoid overflowing the stack on stock kernels with 8k stacks. See #1778 for additional details. Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Ned Bass <bass6@llnl.gov> Closes #1778
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
return (err);
2008-11-20 20:01:55 +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
{
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,
uint64_t txg_start, zbookmark_phys_t *resume, int flags,
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
}
/*
* NB: pool must not be changing on-disk (eg, from zdb or sync context).
*/
int
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
{
int err;
uint64_t obj;
dsl_pool_t *dp = spa_get_dsl(spa);
objset_t *mos = dp->dp_meta_objset;
boolean_t hard = (flags & TRAVERSE_HARD);
/* visit the MOS */
err = traverse_impl(spa, NULL, 0, spa_get_rootblkptr(spa),
txg_start, NULL, flags, func, arg);
if (err != 0)
return (err);
/* visit each dataset */
for (obj = 1; err == 0;
err = dmu_object_next(mos, &obj, FALSE, txg_start)) {
dmu_object_info_t doi;
err = dmu_object_info(mos, obj, &doi);
if (err != 0) {
if (hard)
continue;
break;
}
if (doi.doi_bonus_type == DMU_OT_DSL_DATASET) {
dsl_dataset_t *ds;
uint64_t txg = txg_start;
dsl_pool_config_enter(dp, FTAG);
err = dsl_dataset_hold_obj(dp, obj, FTAG, &ds);
dsl_pool_config_exit(dp, FTAG);
if (err != 0) {
if (hard)
continue;
break;
}
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);
dsl_dataset_rele(ds, FTAG);
if (err != 0)
break;
}
2008-11-20 20:01:55 +00:00
}
if (err == ESRCH)
err = 0;
return (err);
2008-11-20 20:01:55 +00:00
}
#if defined(_KERNEL) && defined(HAVE_SPL)
EXPORT_SYMBOL(traverse_dataset);
EXPORT_SYMBOL(traverse_pool);
Add missing ZFS tunables This commit adds module options for all existing zfs tunables. Ideally the average user should never need to modify any of these values. However, in practice sometimes you do need to tweak these values for one reason or another. In those cases it's nice not to have to resort to rebuilding from source. All tunables are visable to modinfo and the list is as follows: $ modinfo module/zfs/zfs.ko filename: module/zfs/zfs.ko license: CDDL author: Sun Microsystems/Oracle, Lawrence Livermore National Laboratory description: ZFS srcversion: 8EAB1D71DACE05B5AA61567 depends: spl,znvpair,zcommon,zunicode,zavl vermagic: 2.6.32-131.0.5.el6.x86_64 SMP mod_unload modversions parm: zvol_major:Major number for zvol device (uint) parm: zvol_threads:Number of threads for zvol device (uint) parm: zio_injection_enabled:Enable fault injection (int) parm: zio_bulk_flags:Additional flags to pass to bulk buffers (int) parm: zio_delay_max:Max zio millisec delay before posting event (int) parm: zio_requeue_io_start_cut_in_line:Prioritize requeued I/O (bool) parm: zil_replay_disable:Disable intent logging replay (int) parm: zfs_nocacheflush:Disable cache flushes (bool) parm: zfs_read_chunk_size:Bytes to read per chunk (long) parm: zfs_vdev_max_pending:Max pending per-vdev I/Os (int) parm: zfs_vdev_min_pending:Min pending per-vdev I/Os (int) parm: zfs_vdev_aggregation_limit:Max vdev I/O aggregation size (int) parm: zfs_vdev_time_shift:Deadline time shift for vdev I/O (int) parm: zfs_vdev_ramp_rate:Exponential I/O issue ramp-up rate (int) parm: zfs_vdev_read_gap_limit:Aggregate read I/O over gap (int) parm: zfs_vdev_write_gap_limit:Aggregate write I/O over gap (int) parm: zfs_vdev_scheduler:I/O scheduler (charp) parm: zfs_vdev_cache_max:Inflate reads small than max (int) parm: zfs_vdev_cache_size:Total size of the per-disk cache (int) parm: zfs_vdev_cache_bshift:Shift size to inflate reads too (int) parm: zfs_scrub_limit:Max scrub/resilver I/O per leaf vdev (int) parm: zfs_recover:Set to attempt to recover from fatal errors (int) parm: spa_config_path:SPA config file (/etc/zfs/zpool.cache) (charp) parm: zfs_zevent_len_max:Max event queue length (int) parm: zfs_zevent_cols:Max event column width (int) parm: zfs_zevent_console:Log events to the console (int) parm: zfs_top_maxinflight:Max I/Os per top-level (int) parm: zfs_resilver_delay:Number of ticks to delay resilver (int) parm: zfs_scrub_delay:Number of ticks to delay scrub (int) parm: zfs_scan_idle:Idle window in clock ticks (int) parm: zfs_scan_min_time_ms:Min millisecs to scrub per txg (int) parm: zfs_free_min_time_ms:Min millisecs to free per txg (int) parm: zfs_resilver_min_time_ms:Min millisecs to resilver per txg (int) parm: zfs_no_scrub_io:Set to disable scrub I/O (bool) parm: zfs_no_scrub_prefetch:Set to disable scrub prefetching (bool) parm: zfs_txg_timeout:Max seconds worth of delta per txg (int) parm: zfs_no_write_throttle:Disable write throttling (int) parm: zfs_write_limit_shift:log2(fraction of memory) per txg (int) parm: zfs_txg_synctime_ms:Target milliseconds between tgx sync (int) parm: zfs_write_limit_min:Min tgx write limit (ulong) parm: zfs_write_limit_max:Max tgx write limit (ulong) parm: zfs_write_limit_inflated:Inflated tgx write limit (ulong) parm: zfs_write_limit_override:Override tgx write limit (ulong) parm: zfs_prefetch_disable:Disable all ZFS prefetching (int) parm: zfetch_max_streams:Max number of streams per zfetch (uint) parm: zfetch_min_sec_reap:Min time before stream reclaim (uint) parm: zfetch_block_cap:Max number of blocks to fetch at a time (uint) parm: zfetch_array_rd_sz:Number of bytes in a array_read (ulong) parm: zfs_pd_blks_max:Max number of blocks to prefetch (int) parm: zfs_dedup_prefetch:Enable prefetching dedup-ed blks (int) parm: zfs_arc_min:Min arc size (ulong) parm: zfs_arc_max:Max arc size (ulong) parm: zfs_arc_meta_limit:Meta limit for arc size (ulong) parm: zfs_arc_reduce_dnlc_percent:Meta reclaim percentage (int) parm: zfs_arc_grow_retry:Seconds before growing arc size (int) parm: zfs_arc_shrink_shift:log2(fraction of arc to reclaim) (int) parm: zfs_arc_p_min_shift:arc_c shift to calc min/max arc_p (int)
2011-05-03 22:09:28 +00:00
module_param(zfs_pd_blks_max, int, 0644);
MODULE_PARM_DESC(zfs_pd_blks_max, "Max number of blocks to prefetch");
#endif