/*
* Read a log block, make sure it's valid, and byteswap it if necessary.
*/
static int
zil_read_log_block(zilog_t *zilog, const blkptr_t *bp, arc_buf_t **abufpp)
{
blkptr_t blk = *bp;
zbookmark_t zb;
uint32_t aflags = ARC_WAIT;
int error;
zb.zb_objset = bp->blk_cksum.zc_word[ZIL_ZC_OBJSET];
zb.zb_object = 0;
zb.zb_level = -1;
zb.zb_blkid = bp->blk_cksum.zc_word[ZIL_ZC_SEQ];
*abufpp = NULL;
/*
* We shouldn't be doing any scrubbing while we're doing log
* replay, it's OK to not lock.
*/
error = arc_read_nolock(NULL, zilog->zl_spa, &blk,
arc_getbuf_func, abufpp, ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL |
ZIO_FLAG_SPECULATIVE | ZIO_FLAG_SCRUB, &aflags, &zb);
if (error == 0) {
char *data = (*abufpp)->b_data;
uint64_t blksz = BP_GET_LSIZE(bp);
zil_trailer_t *ztp = (zil_trailer_t *)(data + blksz) - 1;
zio_cksum_t cksum = bp->blk_cksum;
/*
* Validate the checksummed log block.
*
* Sequence numbers should be... sequential. The checksum
* verifier for the next block should be bp's checksum plus 1.
*
* Also check the log chain linkage and size used.
*/
cksum.zc_word[ZIL_ZC_SEQ]++;
if (bcmp(&cksum, &ztp->zit_next_blk.blk_cksum,
sizeof (cksum)) || BP_IS_HOLE(&ztp->zit_next_blk) ||
(ztp->zit_nused > (blksz - sizeof (zil_trailer_t)))) {
error = ECKSUM;
}
if (error) {
VERIFY(arc_buf_remove_ref(*abufpp, abufpp) == 1);
*abufpp = NULL;
}
}
dprintf("error %d on %llu:%llu\n", error, zb.zb_objset, zb.zb_blkid);
return (error);
}
void
dmu_objset_evict(dsl_dataset_t *ds, void *arg)
{
objset_impl_t *osi = arg;
objset_t os;
int i;
for (i = 0; i < TXG_SIZE; i++) {
ASSERT(list_head(&osi->os_dirty_dnodes[i]) == NULL);
ASSERT(list_head(&osi->os_free_dnodes[i]) == NULL);
}
if (ds) {
if (!dsl_dataset_is_snapshot(ds)) {
VERIFY(0 == dsl_prop_unregister(ds, "checksum",
checksum_changed_cb, osi));
VERIFY(0 == dsl_prop_unregister(ds, "compression",
compression_changed_cb, osi));
VERIFY(0 == dsl_prop_unregister(ds, "copies",
copies_changed_cb, osi));
}
VERIFY(0 == dsl_prop_unregister(ds, "primarycache",
primary_cache_changed_cb, osi));
VERIFY(0 == dsl_prop_unregister(ds, "secondarycache",
secondary_cache_changed_cb, osi));
}
/*
* We should need only a single pass over the dnode list, since
* nothing can be added to the list at this point.
*/
os.os = osi;
(void) dmu_objset_evict_dbufs(&os);
dnode_special_close(osi->os_meta_dnode);
if (osi->os_userused_dnode) {
dnode_special_close(osi->os_userused_dnode);
dnode_special_close(osi->os_groupused_dnode);
}
zil_free(osi->os_zil);
ASSERT3P(list_head(&osi->os_dnodes), ==, NULL);
VERIFY(arc_buf_remove_ref(osi->os_phys_buf, &osi->os_phys_buf) == 1);
mutex_destroy(&osi->os_lock);
mutex_destroy(&osi->os_obj_lock);
mutex_destroy(&osi->os_user_ptr_lock);
kmem_free(osi, sizeof (objset_impl_t));
}
/*
* return true if the initial log block is not valid
*/
static boolean_t
zil_empty(zilog_t *zilog)
{
const zil_header_t *zh = zilog->zl_header;
arc_buf_t *abuf = NULL;
if (BP_IS_HOLE(&zh->zh_log))
return (B_TRUE);
if (zil_read_log_block(zilog, &zh->zh_log, &abuf) != 0)
return (B_TRUE);
VERIFY(arc_buf_remove_ref(abuf, &abuf) == 1);
return (B_FALSE);
}
/* ARGSUSED */
int
zil_check_log_chain(char *osname, void *txarg)
{
zilog_t *zilog;
zil_header_t *zh;
blkptr_t blk;
arc_buf_t *abuf;
objset_t *os;
char *lrbuf;
zil_trailer_t *ztp;
int error;
error = dmu_objset_open(osname, DMU_OST_ANY, DS_MODE_USER, &os);
if (error) {
cmn_err(CE_WARN, "can't open objset for %s", osname);
return (0);
}
zilog = dmu_objset_zil(os);
zh = zil_header_in_syncing_context(zilog);
blk = zh->zh_log;
if (BP_IS_HOLE(&blk)) {
dmu_objset_close(os);
return (0); /* no chain */
}
for (;;) {
error = zil_read_log_block(zilog, &blk, &abuf);
if (error)
break;
lrbuf = abuf->b_data;
ztp = (zil_trailer_t *)(lrbuf + BP_GET_LSIZE(&blk)) - 1;
blk = ztp->zit_next_blk;
VERIFY(arc_buf_remove_ref(abuf, &abuf) == 1);
}
dmu_objset_close(os);
if (error == ECKSUM)
return (0); /* normal end of chain */
return (error);
}
int
dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
objset_t **osp)
{
objset_t *os;
int i, err;
ASSERT(ds == NULL || MUTEX_HELD(&ds->ds_opening_lock));
os = kmem_zalloc(sizeof (objset_t), KM_SLEEP);
os->os_dsl_dataset = ds;
os->os_spa = spa;
os->os_rootbp = bp;
if (!BP_IS_HOLE(os->os_rootbp)) {
arc_flags_t aflags = ARC_FLAG_WAIT;
zbookmark_phys_t zb;
SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET,
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
if (DMU_OS_IS_L2CACHEABLE(os))
aflags |= ARC_FLAG_L2CACHE;
if (DMU_OS_IS_L2COMPRESSIBLE(os))
aflags |= ARC_FLAG_L2COMPRESS;
dprintf_bp(os->os_rootbp, "reading %s", "");
err = arc_read(NULL, spa, os->os_rootbp,
arc_getbuf_func, &os->os_phys_buf,
ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &aflags, &zb);
if (err != 0) {
kmem_free(os, sizeof (objset_t));
/* convert checksum errors into IO errors */
if (err == ECKSUM)
err = SET_ERROR(EIO);
return (err);
}
/* Increase the blocksize if we are permitted. */
if (spa_version(spa) >= SPA_VERSION_USERSPACE &&
arc_buf_size(os->os_phys_buf) < sizeof (objset_phys_t)) {
arc_buf_t *buf = arc_buf_alloc(spa,
sizeof (objset_phys_t), &os->os_phys_buf,
ARC_BUFC_METADATA);
bzero(buf->b_data, sizeof (objset_phys_t));
bcopy(os->os_phys_buf->b_data, buf->b_data,
arc_buf_size(os->os_phys_buf));
(void) arc_buf_remove_ref(os->os_phys_buf,
&os->os_phys_buf);
os->os_phys_buf = buf;
}
os->os_phys = os->os_phys_buf->b_data;
os->os_flags = os->os_phys->os_flags;
} else {
int size = spa_version(spa) >= SPA_VERSION_USERSPACE ?
sizeof (objset_phys_t) : OBJSET_OLD_PHYS_SIZE;
os->os_phys_buf = arc_buf_alloc(spa, size,
&os->os_phys_buf, ARC_BUFC_METADATA);
os->os_phys = os->os_phys_buf->b_data;
bzero(os->os_phys, size);
}
/*
* Note: the changed_cb will be called once before the register
* func returns, thus changing the checksum/compression from the
* default (fletcher2/off). Snapshots don't need to know about
* checksum/compression/copies.
*/
if (ds != NULL) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_PRIMARYCACHE),
primary_cache_changed_cb, os);
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_SECONDARYCACHE),
secondary_cache_changed_cb, os);
}
if (!ds->ds_is_snapshot) {
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_CHECKSUM),
checksum_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_COMPRESSION),
compression_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_COPIES),
copies_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_DEDUP),
dedup_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_LOGBIAS),
logbias_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_SYNC),
sync_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(
ZFS_PROP_REDUNDANT_METADATA),
redundant_metadata_changed_cb, os);
}
if (err == 0) {
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_RECORDSIZE),
recordsize_changed_cb, os);
}
}
if (err != 0) {
VERIFY(arc_buf_remove_ref(os->os_phys_buf,
&os->os_phys_buf));
kmem_free(os, sizeof (objset_t));
return (err);
}
} else {
/* It's the meta-objset. */
os->os_checksum = ZIO_CHECKSUM_FLETCHER_4;
os->os_compress = ZIO_COMPRESS_ON;
os->os_copies = spa_max_replication(spa);
os->os_dedup_checksum = ZIO_CHECKSUM_OFF;
os->os_dedup_verify = B_FALSE;
os->os_logbias = ZFS_LOGBIAS_LATENCY;
os->os_sync = ZFS_SYNC_STANDARD;
os->os_primary_cache = ZFS_CACHE_ALL;
os->os_secondary_cache = ZFS_CACHE_ALL;
}
if (ds == NULL || !ds->ds_is_snapshot)
os->os_zil_header = os->os_phys->os_zil_header;
os->os_zil = zil_alloc(os, &os->os_zil_header);
for (i = 0; i < TXG_SIZE; i++) {
list_create(&os->os_dirty_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
list_create(&os->os_free_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
}
list_create(&os->os_dnodes, sizeof (dnode_t),
offsetof(dnode_t, dn_link));
list_create(&os->os_downgraded_dbufs, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_link));
mutex_init(&os->os_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&os->os_obj_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&os->os_user_ptr_lock, NULL, MUTEX_DEFAULT, NULL);
dnode_special_open(os, &os->os_phys->os_meta_dnode,
DMU_META_DNODE_OBJECT, &os->os_meta_dnode);
if (arc_buf_size(os->os_phys_buf) >= sizeof (objset_phys_t)) {
dnode_special_open(os, &os->os_phys->os_userused_dnode,
DMU_USERUSED_OBJECT, &os->os_userused_dnode);
dnode_special_open(os, &os->os_phys->os_groupused_dnode,
DMU_GROUPUSED_OBJECT, &os->os_groupused_dnode);
}
*osp = os;
return (0);
}
void
dmu_objset_evict(objset_t *os)
{
dsl_dataset_t *ds = os->os_dsl_dataset;
int t;
for (t = 0; t < TXG_SIZE; t++)
ASSERT(!dmu_objset_is_dirty(os, t));
if (ds) {
if (!dsl_dataset_is_snapshot(ds)) {
VERIFY(0 == dsl_prop_unregister(ds, "checksum",
checksum_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "compression",
compression_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "copies",
copies_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "dedup",
dedup_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "logbias",
logbias_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "sync",
sync_changed_cb, os));
}
VERIFY(0 == dsl_prop_unregister(ds, "primarycache",
primary_cache_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "secondarycache",
secondary_cache_changed_cb, os));
VERIFY(0 == dsl_prop_unregister(ds, "encryption",
crypt_changed_cb, os));
}
if (os->os_sa)
sa_tear_down(os);
/*
* We should need only a single pass over the dnode list, since
* nothing can be added to the list at this point.
*/
(void) dmu_objset_evict_dbufs(os);
dnode_special_close(&os->os_meta_dnode);
if (DMU_USERUSED_DNODE(os)) {
dnode_special_close(&os->os_userused_dnode);
dnode_special_close(&os->os_groupused_dnode);
}
zil_free(os->os_zil);
ASSERT3P(list_head(&os->os_dnodes), ==, NULL);
VERIFY(arc_buf_remove_ref(os->os_phys_buf, &os->os_phys_buf) == 1);
/*
* This is a barrier to prevent the objset from going away in
* dnode_move() until we can safely ensure that the objset is still in
* use. We consider the objset valid before the barrier and invalid
* after the barrier.
*/
rw_enter(&os_lock, RW_READER);
rw_exit(&os_lock);
mutex_destroy(&os->os_lock);
mutex_destroy(&os->os_obj_lock);
mutex_destroy(&os->os_user_ptr_lock);
kmem_free(os, sizeof (objset_t));
}
int
dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
objset_t **osp)
{
objset_t *os;
int i, err = 0;
ASSERT(ds == NULL || MUTEX_HELD(&ds->ds_opening_lock));
os = kmem_zalloc(sizeof (objset_t), KM_PUSHPAGE);
os->os_dsl_dataset = ds;
os->os_spa = spa;
os->os_rootbp = bp;
if (!BP_IS_HOLE(os->os_rootbp)) {
uint32_t aflags = ARC_WAIT;
zbookmark_t zb;
SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET,
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
if (DMU_OS_IS_L2CACHEABLE(os))
aflags |= ARC_L2CACHE;
dprintf_bp(os->os_rootbp, "reading %s", "");
/*
* XXX when bprewrite scrub can change the bp,
* and this is called from dmu_objset_open_ds_os, the bp
* could change, and we'll need a lock.
*/
err = dsl_read_nolock(NULL, spa, os->os_rootbp,
arc_getbuf_func, &os->os_phys_buf,
ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &aflags, &zb);
if (err) {
kmem_free(os, sizeof (objset_t));
/* convert checksum errors into IO errors */
if (err == ECKSUM)
err = EIO;
return (err);
}
/* Increase the blocksize if we are permitted. */
if (spa_version(spa) >= SPA_VERSION_USERSPACE &&
arc_buf_size(os->os_phys_buf) < sizeof (objset_phys_t)) {
arc_buf_t *buf = arc_buf_alloc(spa,
sizeof (objset_phys_t), &os->os_phys_buf,
ARC_BUFC_METADATA);
bzero(buf->b_data, sizeof (objset_phys_t));
bcopy(os->os_phys_buf->b_data, buf->b_data,
arc_buf_size(os->os_phys_buf));
(void) arc_buf_remove_ref(os->os_phys_buf,
&os->os_phys_buf);
os->os_phys_buf = buf;
}
os->os_phys = os->os_phys_buf->b_data;
os->os_flags = os->os_phys->os_flags;
} else {
int size = spa_version(spa) >= SPA_VERSION_USERSPACE ?
sizeof (objset_phys_t) : OBJSET_OLD_PHYS_SIZE;
os->os_phys_buf = arc_buf_alloc(spa, size,
&os->os_phys_buf, ARC_BUFC_METADATA);
os->os_phys = os->os_phys_buf->b_data;
bzero(os->os_phys, size);
}
/*
* Note: the changed_cb will be called once before the register
* func returns, thus changing the checksum/compression from the
* default (fletcher2/off). Snapshots don't need to know about
* checksum/compression/copies. But they do need to know about
* encryption so that clones from the snaphost inherit the
* same encryption property regardless of where in the namespace
* they get created.
*/
if (ds) {
err = dsl_prop_register(ds, "primarycache",
primary_cache_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "secondarycache",
secondary_cache_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "encryption",
crypt_changed_cb, os);
if (!dsl_dataset_is_snapshot(ds)) {
if (err == 0)
err = dsl_prop_register(ds, "checksum",
checksum_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "compression",
compression_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "copies",
copies_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "dedup",
dedup_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "logbias",
logbias_changed_cb, os);
if (err == 0)
err = dsl_prop_register(ds, "sync",
sync_changed_cb, os);
}
if (err) {
VERIFY(arc_buf_remove_ref(os->os_phys_buf,
&os->os_phys_buf) == 1);
kmem_free(os, sizeof (objset_t));
return (err);
}
} else if (ds == NULL) {
/*
* It's the meta-objset.
* Encryption is off for ZFS metadata but on for ZPL metadata
* and file/zvol contents.
*/
os->os_checksum = ZIO_CHECKSUM_FLETCHER_4;
os->os_compress = ZIO_COMPRESS_LZJB;
os->os_copies = spa_max_replication(spa);
os->os_dedup_checksum = ZIO_CHECKSUM_OFF;
os->os_dedup_verify = 0;
os->os_logbias = 0;
os->os_sync = 0;
os->os_primary_cache = ZFS_CACHE_ALL;
os->os_secondary_cache = ZFS_CACHE_ALL;
os->os_crypt = ZIO_CRYPT_OFF;
}
if (ds == NULL || !dsl_dataset_is_snapshot(ds))
os->os_zil_header = os->os_phys->os_zil_header;
os->os_zil = zil_alloc(os, &os->os_zil_header);
for (i = 0; i < TXG_SIZE; i++) {
list_create(&os->os_dirty_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
list_create(&os->os_free_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
}
list_create(&os->os_dnodes, sizeof (dnode_t),
offsetof(dnode_t, dn_link));
list_create(&os->os_downgraded_dbufs, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_link));
mutex_init(&os->os_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&os->os_obj_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&os->os_user_ptr_lock, NULL, MUTEX_DEFAULT, NULL);
DMU_META_DNODE(os) = dnode_special_open(os,
&os->os_phys->os_meta_dnode, DMU_META_DNODE_OBJECT,
&os->os_meta_dnode);
if (arc_buf_size(os->os_phys_buf) >= sizeof (objset_phys_t)) {
DMU_USERUSED_DNODE(os) = dnode_special_open(os,
&os->os_phys->os_userused_dnode, DMU_USERUSED_OBJECT,
&os->os_userused_dnode);
DMU_GROUPUSED_DNODE(os) = dnode_special_open(os,
&os->os_phys->os_groupused_dnode, DMU_GROUPUSED_OBJECT,
&os->os_groupused_dnode);
}
/*
* We should be the only thread trying to do this because we
* have ds_opening_lock
*/
if (ds) {
mutex_enter(&ds->ds_lock);
ASSERT(ds->ds_objset == NULL);
ds->ds_objset = os;
mutex_exit(&ds->ds_lock);
}
*osp = os;
return (0);
}
void
dmu_objset_evict(objset_t *os)
{
dsl_dataset_t *ds = os->os_dsl_dataset;
for (int t = 0; t < TXG_SIZE; t++)
ASSERT(!dmu_objset_is_dirty(os, t));
if (ds) {
if (!dsl_dataset_is_snapshot(ds)) {
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_CHECKSUM),
checksum_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_COMPRESSION),
compression_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_COPIES),
copies_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_DEDUP),
dedup_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_LOGBIAS),
logbias_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_SYNC),
sync_changed_cb, os));
}
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_PRIMARYCACHE),
primary_cache_changed_cb, os));
VERIFY0(dsl_prop_unregister(ds,
zfs_prop_to_name(ZFS_PROP_SECONDARYCACHE),
secondary_cache_changed_cb, os));
}
if (os->os_sa)
sa_tear_down(os);
dmu_objset_evict_dbufs(os);
dnode_special_close(&os->os_meta_dnode);
if (DMU_USERUSED_DNODE(os)) {
dnode_special_close(&os->os_userused_dnode);
dnode_special_close(&os->os_groupused_dnode);
}
zil_free(os->os_zil);
ASSERT3P(list_head(&os->os_dnodes), ==, NULL);
VERIFY(arc_buf_remove_ref(os->os_phys_buf, &os->os_phys_buf));
/*
* This is a barrier to prevent the objset from going away in
* dnode_move() until we can safely ensure that the objset is still in
* use. We consider the objset valid before the barrier and invalid
* after the barrier.
*/
rw_enter(&os_lock, RW_READER);
rw_exit(&os_lock);
mutex_destroy(&os->os_lock);
mutex_destroy(&os->os_obj_lock);
mutex_destroy(&os->os_user_ptr_lock);
kmem_free(os, sizeof (objset_t));
}
int
dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
objset_impl_t **osip)
{
objset_impl_t *winner, *osi;
int i, err, checksum;
osi = kmem_zalloc(sizeof (objset_impl_t), KM_SLEEP);
osi->os.os = osi;
osi->os_dsl_dataset = ds;
osi->os_spa = spa;
osi->os_rootbp = bp;
if (!BP_IS_HOLE(osi->os_rootbp)) {
uint32_t aflags = ARC_WAIT;
zbookmark_t zb;
zb.zb_objset = ds ? ds->ds_object : 0;
zb.zb_object = 0;
zb.zb_level = -1;
zb.zb_blkid = 0;
dprintf_bp(osi->os_rootbp, "reading %s", "");
err = arc_read(NULL, spa, osi->os_rootbp,
dmu_ot[DMU_OT_OBJSET].ot_byteswap,
arc_getbuf_func, &osi->os_phys_buf,
ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &aflags, &zb);
if (err) {
kmem_free(osi, sizeof (objset_impl_t));
return (err);
}
osi->os_phys = osi->os_phys_buf->b_data;
arc_release(osi->os_phys_buf, &osi->os_phys_buf);
} else {
osi->os_phys_buf = arc_buf_alloc(spa, sizeof (objset_phys_t),
&osi->os_phys_buf, ARC_BUFC_METADATA);
osi->os_phys = osi->os_phys_buf->b_data;
bzero(osi->os_phys, sizeof (objset_phys_t));
}
/*
* Note: the changed_cb will be called once before the register
* func returns, thus changing the checksum/compression from the
* default (fletcher2/off). Snapshots don't need to know, and
* registering would complicate clone promotion.
*/
if (ds && ds->ds_phys->ds_num_children == 0) {
err = dsl_prop_register(ds, "checksum",
checksum_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "compression",
compression_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "copies",
copies_changed_cb, osi);
if (err) {
VERIFY(arc_buf_remove_ref(osi->os_phys_buf,
&osi->os_phys_buf) == 1);
kmem_free(osi, sizeof (objset_impl_t));
return (err);
}
} else if (ds == NULL) {
/* It's the meta-objset. */
osi->os_checksum = ZIO_CHECKSUM_FLETCHER_4;
osi->os_compress = ZIO_COMPRESS_LZJB;
osi->os_copies = spa_max_replication(spa);
}
osi->os_zil = zil_alloc(&osi->os, &osi->os_phys->os_zil_header);
/*
* Metadata always gets compressed and checksummed.
* If the data checksum is multi-bit correctable, and it's not
* a ZBT-style checksum, then it's suitable for metadata as well.
* Otherwise, the metadata checksum defaults to fletcher4.
*/
checksum = osi->os_checksum;
if (zio_checksum_table[checksum].ci_correctable &&
!zio_checksum_table[checksum].ci_zbt)
osi->os_md_checksum = checksum;
else
osi->os_md_checksum = ZIO_CHECKSUM_FLETCHER_4;
osi->os_md_compress = ZIO_COMPRESS_LZJB;
for (i = 0; i < TXG_SIZE; i++) {
list_create(&osi->os_dirty_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
list_create(&osi->os_free_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
}
list_create(&osi->os_dnodes, sizeof (dnode_t),
offsetof(dnode_t, dn_link));
list_create(&osi->os_downgraded_dbufs, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_link));
mutex_init(&osi->os_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&osi->os_obj_lock, NULL, MUTEX_DEFAULT, NULL);
osi->os_meta_dnode = dnode_special_open(osi,
&osi->os_phys->os_meta_dnode, DMU_META_DNODE_OBJECT);
if (ds != NULL) {
winner = dsl_dataset_set_user_ptr(ds, osi, dmu_objset_evict);
if (winner) {
dmu_objset_evict(ds, osi);
osi = winner;
}
}
*osip = osi;
return (0);
}
int
dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
objset_impl_t **osip)
{
objset_impl_t *osi;
int i, err;
ASSERT(ds == NULL || MUTEX_HELD(&ds->ds_opening_lock));
osi = kmem_zalloc(sizeof (objset_impl_t), KM_SLEEP);
osi->os.os = osi;
osi->os_dsl_dataset = ds;
osi->os_spa = spa;
osi->os_rootbp = bp;
if (!BP_IS_HOLE(osi->os_rootbp)) {
uint32_t aflags = ARC_WAIT;
zbookmark_t zb;
zb.zb_objset = ds ? ds->ds_object : 0;
zb.zb_object = 0;
zb.zb_level = -1;
zb.zb_blkid = 0;
dprintf_bp(osi->os_rootbp, "reading %s", "");
/*
* NB: when bprewrite scrub can change the bp,
* and this is called from dmu_objset_open_ds_os, the bp
* could change, and we'll need a lock.
*/
err = arc_read_nolock(NULL, spa, osi->os_rootbp,
arc_getbuf_func, &osi->os_phys_buf,
ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &aflags, &zb);
if (err) {
kmem_free(osi, sizeof (objset_impl_t));
return (err);
}
osi->os_phys = osi->os_phys_buf->b_data;
} else {
#ifdef __APPLE_KERNEL__
osi->os_phys_buf = arc_buf_alloc(spa, sizeof (objset_phys_t),
&osi->os_phys_buf, ARC_BUFC_METADATA, TRUE/*alloc_buf*/);
#else
osi->os_phys_buf = arc_buf_alloc(spa, sizeof (objset_phys_t),
&osi->os_phys_buf, ARC_BUFC_METADATA);
#endif
osi->os_phys = osi->os_phys_buf->b_data;
bzero(osi->os_phys, sizeof (objset_phys_t));
}
/*
* Note: the changed_cb will be called once before the register
* func returns, thus changing the checksum/compression from the
* default (fletcher2/off). Snapshots don't need to know, and
* registering would complicate clone promotion.
*/
if (ds && ds->ds_phys->ds_num_children == 0) {
err = dsl_prop_register(ds, "checksum",
checksum_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "compression",
compression_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "copies",
copies_changed_cb, osi);
if (err) {
VERIFY(arc_buf_remove_ref(osi->os_phys_buf,
&osi->os_phys_buf) == 1);
kmem_free(osi, sizeof (objset_impl_t));
return (err);
}
} else if (ds == NULL) {
/* It's the meta-objset. */
osi->os_checksum = ZIO_CHECKSUM_FLETCHER_4;
osi->os_compress = ZIO_COMPRESS_LZJB;
osi->os_copies = spa_max_replication(spa);
}
osi->os_zil_header = osi->os_phys->os_zil_header;
osi->os_zil = zil_alloc(&osi->os, &osi->os_zil_header);
for (i = 0; i < TXG_SIZE; i++) {
list_create(&osi->os_dirty_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
list_create(&osi->os_free_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
}
list_create(&osi->os_dnodes, sizeof (dnode_t),
offsetof(dnode_t, dn_link));
list_create(&osi->os_downgraded_dbufs, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_link));
mutex_init(&osi->os_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&osi->os_obj_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&osi->os_user_ptr_lock, NULL, MUTEX_DEFAULT, NULL);
osi->os_meta_dnode = dnode_special_open(osi,
&osi->os_phys->os_meta_dnode, DMU_META_DNODE_OBJECT);
/*
* We should be the only thread trying to do this because we
* have ds_opening_lock
*/
if (ds) {
VERIFY(NULL == dsl_dataset_set_user_ptr(ds, osi,
dmu_objset_evict));
}
*osip = osi;
return (0);
}
int
dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
objset_impl_t **osip)
{
objset_impl_t *osi;
int i, err;
ASSERT(ds == NULL || MUTEX_HELD(&ds->ds_opening_lock));
osi = kmem_zalloc(sizeof (objset_impl_t), KM_SLEEP);
osi->os.os = osi;
osi->os_dsl_dataset = ds;
osi->os_spa = spa;
osi->os_rootbp = bp;
if (!BP_IS_HOLE(osi->os_rootbp)) {
uint32_t aflags = ARC_WAIT;
zbookmark_t zb;
zb.zb_objset = ds ? ds->ds_object : 0;
zb.zb_object = 0;
zb.zb_level = -1;
zb.zb_blkid = 0;
if (DMU_OS_IS_L2CACHEABLE(osi))
aflags |= ARC_L2CACHE;
dprintf_bp(osi->os_rootbp, "reading %s", "");
/*
* NB: when bprewrite scrub can change the bp,
* and this is called from dmu_objset_open_ds_os, the bp
* could change, and we'll need a lock.
*/
err = arc_read_nolock(NULL, spa, osi->os_rootbp,
arc_getbuf_func, &osi->os_phys_buf,
ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL, &aflags, &zb);
if (err) {
kmem_free(osi, sizeof (objset_impl_t));
/* convert checksum errors into IO errors */
if (err == ECKSUM)
err = EIO;
return (err);
}
/* Increase the blocksize if we are permitted. */
if (spa_version(spa) >= SPA_VERSION_USERSPACE &&
arc_buf_size(osi->os_phys_buf) < sizeof (objset_phys_t)) {
arc_buf_t *buf = arc_buf_alloc(spa,
sizeof (objset_phys_t), &osi->os_phys_buf,
ARC_BUFC_METADATA);
bzero(buf->b_data, sizeof (objset_phys_t));
bcopy(osi->os_phys_buf->b_data, buf->b_data,
arc_buf_size(osi->os_phys_buf));
(void) arc_buf_remove_ref(osi->os_phys_buf,
&osi->os_phys_buf);
osi->os_phys_buf = buf;
}
osi->os_phys = osi->os_phys_buf->b_data;
osi->os_flags = osi->os_phys->os_flags;
} else {
int size = spa_version(spa) >= SPA_VERSION_USERSPACE ?
sizeof (objset_phys_t) : OBJSET_OLD_PHYS_SIZE;
osi->os_phys_buf = arc_buf_alloc(spa, size,
&osi->os_phys_buf, ARC_BUFC_METADATA);
osi->os_phys = osi->os_phys_buf->b_data;
bzero(osi->os_phys, size);
}
/*
* Note: the changed_cb will be called once before the register
* func returns, thus changing the checksum/compression from the
* default (fletcher2/off). Snapshots don't need to know about
* checksum/compression/copies.
*/
if (ds) {
err = dsl_prop_register(ds, "primarycache",
primary_cache_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "secondarycache",
secondary_cache_changed_cb, osi);
if (!dsl_dataset_is_snapshot(ds)) {
if (err == 0)
err = dsl_prop_register(ds, "checksum",
checksum_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "compression",
compression_changed_cb, osi);
if (err == 0)
err = dsl_prop_register(ds, "copies",
copies_changed_cb, osi);
}
if (err) {
VERIFY(arc_buf_remove_ref(osi->os_phys_buf,
&osi->os_phys_buf) == 1);
kmem_free(osi, sizeof (objset_impl_t));
return (err);
}
} else if (ds == NULL) {
/* It's the meta-objset. */
osi->os_checksum = ZIO_CHECKSUM_FLETCHER_4;
osi->os_compress = ZIO_COMPRESS_LZJB;
osi->os_copies = spa_max_replication(spa);
osi->os_primary_cache = ZFS_CACHE_ALL;
osi->os_secondary_cache = ZFS_CACHE_ALL;
}
osi->os_zil_header = osi->os_phys->os_zil_header;
osi->os_zil = zil_alloc(&osi->os, &osi->os_zil_header);
for (i = 0; i < TXG_SIZE; i++) {
list_create(&osi->os_dirty_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
list_create(&osi->os_free_dnodes[i], sizeof (dnode_t),
offsetof(dnode_t, dn_dirty_link[i]));
}
list_create(&osi->os_dnodes, sizeof (dnode_t),
offsetof(dnode_t, dn_link));
list_create(&osi->os_downgraded_dbufs, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_link));
mutex_init(&osi->os_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&osi->os_obj_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&osi->os_user_ptr_lock, NULL, MUTEX_DEFAULT, NULL);
osi->os_meta_dnode = dnode_special_open(osi,
&osi->os_phys->os_meta_dnode, DMU_META_DNODE_OBJECT);
if (arc_buf_size(osi->os_phys_buf) >= sizeof (objset_phys_t)) {
osi->os_userused_dnode = dnode_special_open(osi,
&osi->os_phys->os_userused_dnode, DMU_USERUSED_OBJECT);
osi->os_groupused_dnode = dnode_special_open(osi,
&osi->os_phys->os_groupused_dnode, DMU_GROUPUSED_OBJECT);
}
/*
* We should be the only thread trying to do this because we
* have ds_opening_lock
*/
if (ds) {
VERIFY(NULL == dsl_dataset_set_user_ptr(ds, osi,
dmu_objset_evict));
}
*osip = osi;
return (0);
}
/*
* Parse the intent log, and call parse_func for each valid record within.
* Return the highest sequence number.
*/
uint64_t
zil_parse(zilog_t *zilog, zil_parse_blk_func_t *parse_blk_func,
zil_parse_lr_func_t *parse_lr_func, void *arg, uint64_t txg)
{
const zil_header_t *zh = zilog->zl_header;
uint64_t claim_seq = zh->zh_claim_seq;
uint64_t seq = 0;
uint64_t max_seq = 0;
blkptr_t blk = zh->zh_log;
arc_buf_t *abuf;
char *lrbuf, *lrp;
zil_trailer_t *ztp;
int reclen, error;
if (BP_IS_HOLE(&blk))
return (max_seq);
/*
* Starting at the block pointed to by zh_log we read the log chain.
* For each block in the chain we strongly check that block to
* ensure its validity. We stop when an invalid block is found.
* For each block pointer in the chain we call parse_blk_func().
* For each record in each valid block we call parse_lr_func().
* If the log has been claimed, stop if we encounter a sequence
* number greater than the highest claimed sequence number.
*/
zil_dva_tree_init(&zilog->zl_dva_tree);
for (;;) {
seq = blk.blk_cksum.zc_word[ZIL_ZC_SEQ];
if (claim_seq != 0 && seq > claim_seq)
break;
ASSERT(max_seq < seq);
max_seq = seq;
error = zil_read_log_block(zilog, &blk, &abuf);
if (parse_blk_func != NULL)
parse_blk_func(zilog, &blk, arg, txg);
if (error)
break;
lrbuf = abuf->b_data;
ztp = (zil_trailer_t *)(lrbuf + BP_GET_LSIZE(&blk)) - 1;
blk = ztp->zit_next_blk;
if (parse_lr_func == NULL) {
VERIFY(arc_buf_remove_ref(abuf, &abuf) == 1);
continue;
}
for (lrp = lrbuf; lrp < lrbuf + ztp->zit_nused; lrp += reclen) {
lr_t *lr = (lr_t *)lrp;
reclen = lr->lrc_reclen;
ASSERT3U(reclen, >=, sizeof (lr_t));
parse_lr_func(zilog, lr, arg, txg);
}
VERIFY(arc_buf_remove_ref(abuf, &abuf) == 1);
}
zil_dva_tree_fini(&zilog->zl_dva_tree);
return (max_seq);
}
