* CDDL HEADER END
*/
/*
- * Copyright 2008 Sun Microsystems, Inc. All rights reserved.
- * Use is subject to license terms.
+ * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
*/
#include <sys/zfs_context.h>
}
if (dn->dn_phys->dn_type != DMU_OT_NONE)
ASSERT3U(dn->dn_phys->dn_nlevels, <=, dn->dn_nlevels);
- ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT || dn->dn_dbuf != NULL);
+ ASSERT(DMU_OBJECT_IS_SPECIAL(dn->dn_object) || dn->dn_dbuf != NULL);
if (dn->dn_dbuf != NULL) {
ASSERT3P(dn->dn_phys, ==,
(dnode_phys_t *)dn->dn_dbuf->db.db_data +
ASSERT3U(dnp->dn_bonustype, <, DMU_OT_NUMTYPES);
dmu_ot[dnp->dn_bonustype].ot_byteswap(dnp->dn_bonus + off, len);
}
+
+ /* Swap SPILL block if we have one */
+ if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR)
+ byteswap_uint64_array(&dnp->dn_spill, sizeof (blkptr_t));
+
}
void
rw_exit(&dn->dn_struct_rwlock);
}
+void
+dnode_setbonus_type(dnode_t *dn, dmu_object_type_t newtype, dmu_tx_t *tx)
+{
+ ASSERT3U(refcount_count(&dn->dn_holds), >=, 1);
+ dnode_setdirty(dn, tx);
+ rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
+ dn->dn_bonustype = newtype;
+ dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype;
+ rw_exit(&dn->dn_struct_rwlock);
+}
+
+void
+dnode_rm_spill(dnode_t *dn, dmu_tx_t *tx)
+{
+ ASSERT3U(refcount_count(&dn->dn_holds), >=, 1);
+ ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock));
+ dnode_setdirty(dn, tx);
+ dn->dn_rm_spillblk[tx->tx_txg&TXG_MASK] = DN_KILL_SPILLBLK;
+ dn->dn_have_spill = B_FALSE;
+}
+
static void
dnode_setdblksz(dnode_t *dn, int size)
{
}
static dnode_t *
-dnode_create(objset_impl_t *os, dnode_phys_t *dnp, dmu_buf_impl_t *db,
+dnode_create(objset_t *os, dnode_phys_t *dnp, dmu_buf_impl_t *db,
uint64_t object)
{
dnode_t *dn = kmem_cache_alloc(dnode_cache, KM_SLEEP);
dn->dn_bonustype = dnp->dn_bonustype;
dn->dn_bonuslen = dnp->dn_bonuslen;
dn->dn_maxblkid = dnp->dn_maxblkid;
+ dn->dn_have_spill = ((dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) != 0);
+ dn->dn_id_flags = 0;
dmu_zfetch_init(&dn->dn_zfetch, dn);
list_insert_head(&os->os_dnodes, dn);
mutex_exit(&os->os_lock);
- arc_space_consume(sizeof (dnode_t));
+ arc_space_consume(sizeof (dnode_t), ARC_SPACE_OTHER);
return (dn);
}
static void
dnode_destroy(dnode_t *dn)
{
- objset_impl_t *os = dn->dn_objset;
+ objset_t *os = dn->dn_objset;
#ifdef ZFS_DEBUG
int i;
}
ASSERT(NULL == list_head(&dn->dn_dbufs));
#endif
+ ASSERT((dn->dn_id_flags & DN_ID_NEW_EXIST) == 0);
mutex_enter(&os->os_lock);
list_remove(&os->os_dnodes, dn);
dn->dn_bonus = NULL;
}
kmem_cache_free(dnode_cache, dn);
- arc_space_return(sizeof (dnode_t));
+ arc_space_return(sizeof (dnode_t), ARC_SPACE_OTHER);
}
void
ASSERT(ot != DMU_OT_NONE);
ASSERT3U(ot, <, DMU_OT_NUMTYPES);
ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) ||
+ (bonustype == DMU_OT_SA && bonuslen == 0) ||
(bonustype != DMU_OT_NONE && bonuslen != 0));
ASSERT3U(bonustype, <, DMU_OT_NUMTYPES);
ASSERT3U(bonuslen, <=, DN_MAX_BONUSLEN);
ASSERT3U(dn->dn_next_nlevels[i], ==, 0);
ASSERT3U(dn->dn_next_indblkshift[i], ==, 0);
ASSERT3U(dn->dn_next_bonuslen[i], ==, 0);
+ ASSERT3U(dn->dn_next_bonustype[i], ==, 0);
+ ASSERT3U(dn->dn_rm_spillblk[i], ==, 0);
ASSERT3U(dn->dn_next_blksz[i], ==, 0);
ASSERT(!list_link_active(&dn->dn_dirty_link[i]));
ASSERT3P(list_head(&dn->dn_dirty_records[i]), ==, NULL);
dnode_setdblksz(dn, blocksize);
dn->dn_indblkshift = ibs;
dn->dn_nlevels = 1;
- dn->dn_nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
+ if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */
+ dn->dn_nblkptr = 1;
+ else
+ dn->dn_nblkptr = 1 +
+ ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
dn->dn_bonustype = bonustype;
dn->dn_bonuslen = bonuslen;
dn->dn_checksum = ZIO_CHECKSUM_INHERIT;
}
dn->dn_allocated_txg = tx->tx_txg;
+ dn->dn_id_flags = 0;
dnode_setdirty(dn, tx);
dn->dn_next_indblkshift[tx->tx_txg & TXG_MASK] = ibs;
dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = dn->dn_bonuslen;
+ dn->dn_next_bonustype[tx->tx_txg & TXG_MASK] = dn->dn_bonustype;
dn->dn_next_blksz[tx->tx_txg & TXG_MASK] = dn->dn_datablksz;
}
dnode_reallocate(dnode_t *dn, dmu_object_type_t ot, int blocksize,
dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
{
- int i, old_nblkptr;
- dmu_buf_impl_t *db = NULL;
+ int nblkptr;
ASSERT3U(blocksize, >=, SPA_MINBLOCKSIZE);
ASSERT3U(blocksize, <=, SPA_MAXBLOCKSIZE);
ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx));
ASSERT(tx->tx_txg != 0);
ASSERT((bonustype == DMU_OT_NONE && bonuslen == 0) ||
- (bonustype != DMU_OT_NONE && bonuslen != 0));
+ (bonustype != DMU_OT_NONE && bonuslen != 0) ||
+ (bonustype == DMU_OT_SA && bonuslen == 0));
ASSERT3U(bonustype, <, DMU_OT_NUMTYPES);
ASSERT3U(bonuslen, <=, DN_MAX_BONUSLEN);
- for (i = 0; i < TXG_SIZE; i++)
- ASSERT(!list_link_active(&dn->dn_dirty_link[i]));
-
/* clean up any unreferenced dbufs */
dnode_evict_dbufs(dn);
- ASSERT3P(list_head(&dn->dn_dbufs), ==, NULL);
- /*
- * XXX I should really have a generation number to tell if we
- * need to do this...
- */
- if (blocksize != dn->dn_datablksz ||
- dn->dn_bonustype != bonustype || dn->dn_bonuslen != bonuslen) {
- /* free all old data */
- dnode_free_range(dn, 0, -1ULL, tx);
- }
+ dn->dn_id_flags = 0;
- /* change blocksize */
rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
- if (blocksize != dn->dn_datablksz &&
- (!BP_IS_HOLE(&dn->dn_phys->dn_blkptr[0]) ||
- list_head(&dn->dn_dbufs) != NULL)) {
- db = dbuf_hold(dn, 0, FTAG);
- dbuf_new_size(db, blocksize, tx);
- }
- dnode_setdblksz(dn, blocksize);
dnode_setdirty(dn, tx);
- dn->dn_next_bonuslen[tx->tx_txg&TXG_MASK] = bonuslen;
- dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = blocksize;
+ if (dn->dn_datablksz != blocksize) {
+ /* change blocksize */
+ ASSERT(dn->dn_maxblkid == 0 &&
+ (BP_IS_HOLE(&dn->dn_phys->dn_blkptr[0]) ||
+ dnode_block_freed(dn, 0)));
+ dnode_setdblksz(dn, blocksize);
+ dn->dn_next_blksz[tx->tx_txg&TXG_MASK] = blocksize;
+ }
+ if (dn->dn_bonuslen != bonuslen)
+ dn->dn_next_bonuslen[tx->tx_txg&TXG_MASK] = bonuslen;
+
+ if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */
+ nblkptr = 1;
+ else
+ nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
+ if (dn->dn_bonustype != bonustype)
+ dn->dn_next_bonustype[tx->tx_txg&TXG_MASK] = bonustype;
+ if (dn->dn_nblkptr != nblkptr)
+ dn->dn_next_nblkptr[tx->tx_txg&TXG_MASK] = nblkptr;
+ if (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) {
+ dbuf_rm_spill(dn, tx);
+ dnode_rm_spill(dn, tx);
+ }
rw_exit(&dn->dn_struct_rwlock);
- if (db)
- dbuf_rele(db, FTAG);
/* change type */
dn->dn_type = ot;
/* change bonus size and type */
mutex_enter(&dn->dn_mtx);
- old_nblkptr = dn->dn_nblkptr;
dn->dn_bonustype = bonustype;
dn->dn_bonuslen = bonuslen;
- dn->dn_nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
+ dn->dn_nblkptr = nblkptr;
dn->dn_checksum = ZIO_CHECKSUM_INHERIT;
dn->dn_compress = ZIO_COMPRESS_INHERIT;
ASSERT3U(dn->dn_nblkptr, <=, DN_MAX_NBLKPTR);
- /* XXX - for now, we can't make nblkptr smaller */
- ASSERT3U(dn->dn_nblkptr, >=, old_nblkptr);
-
- /* fix up the bonus db_size if dn_nblkptr has changed */
- if (dn->dn_bonus && dn->dn_bonuslen != old_nblkptr) {
+ /* fix up the bonus db_size */
+ if (dn->dn_bonus) {
dn->dn_bonus->db.db_size =
DN_MAX_BONUSLEN - (dn->dn_nblkptr-1) * sizeof (blkptr_t);
ASSERT(dn->dn_bonuslen <= dn->dn_bonus->db.db_size);
}
dnode_t *
-dnode_special_open(objset_impl_t *os, dnode_phys_t *dnp, uint64_t object)
+dnode_special_open(objset_t *os, dnode_phys_t *dnp, uint64_t object)
{
dnode_t *dn = dnode_create(os, dnp, NULL, object);
DNODE_VERIFY(dn);
* succeeds even for free dnodes.
*/
int
-dnode_hold_impl(objset_impl_t *os, uint64_t object, int flag,
+dnode_hold_impl(objset_t *os, uint64_t object, int flag,
void *tag, dnode_t **dnp)
{
int epb, idx, err;
/*
* If you are holding the spa config lock as writer, you shouldn't
- * be asking the DMU to do *anything*.
+ * be asking the DMU to do *anything* unless it's the root pool
+ * which may require us to read from the root filesystem while
+ * holding some (not all) of the locks as writer.
*/
- ASSERT(spa_config_held(os->os_spa, SCL_ALL, RW_WRITER) == 0);
+ ASSERT(spa_config_held(os->os_spa, SCL_ALL, RW_WRITER) == 0 ||
+ (spa_is_root(os->os_spa) &&
+ spa_config_held(os->os_spa, SCL_STATE, RW_WRITER) &&
+ !spa_config_held(os->os_spa, SCL_ZIO, RW_WRITER)));
+
+ if (object == DMU_USERUSED_OBJECT || object == DMU_GROUPUSED_OBJECT) {
+ dn = (object == DMU_USERUSED_OBJECT) ?
+ os->os_userused_dnode : os->os_groupused_dnode;
+ if (dn == NULL)
+ return (ENOENT);
+ type = dn->dn_type;
+ if ((flag & DNODE_MUST_BE_ALLOCATED) && type == DMU_OT_NONE)
+ return (ENOENT);
+ if ((flag & DNODE_MUST_BE_FREE) && type != DMU_OT_NONE)
+ return (EEXIST);
+ DNODE_VERIFY(dn);
+ (void) refcount_add(&dn->dn_holds, tag);
+ *dnp = dn;
+ return (0);
+ }
if (object == 0 || object >= DN_MAX_OBJECT)
return (EINVAL);
type = dn->dn_type;
if (dn->dn_free_txg ||
((flag & DNODE_MUST_BE_ALLOCATED) && type == DMU_OT_NONE) ||
- ((flag & DNODE_MUST_BE_FREE) && type != DMU_OT_NONE)) {
+ ((flag & DNODE_MUST_BE_FREE) &&
+ (type != DMU_OT_NONE || !refcount_is_zero(&dn->dn_holds)))) {
mutex_exit(&dn->dn_mtx);
dbuf_rele(db, FTAG);
return (type == DMU_OT_NONE ? ENOENT : EEXIST);
* Return held dnode if the object is allocated, NULL if not.
*/
int
-dnode_hold(objset_impl_t *os, uint64_t object, void *tag, dnode_t **dnp)
+dnode_hold(objset_t *os, uint64_t object, void *tag, dnode_t **dnp)
{
return (dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED, tag, dnp));
}
void
dnode_setdirty(dnode_t *dn, dmu_tx_t *tx)
{
- objset_impl_t *os = dn->dn_objset;
+ objset_t *os = dn->dn_objset;
uint64_t txg = tx->tx_txg;
- if (dn->dn_object == DMU_META_DNODE_OBJECT)
+ if (DMU_OBJECT_IS_SPECIAL(dn->dn_object)) {
+ dsl_dataset_dirty(os->os_dsl_dataset, tx);
return;
+ }
DNODE_VERIFY(dn);
mutex_exit(&dn->dn_mtx);
#endif
+ /*
+ * Determine old uid/gid when necessary
+ */
+ dmu_objset_userquota_get_ids(dn, B_TRUE, tx);
+
mutex_enter(&os->os_lock);
/*
ASSERT(dn->dn_datablksz != 0);
ASSERT3U(dn->dn_next_bonuslen[txg&TXG_MASK], ==, 0);
ASSERT3U(dn->dn_next_blksz[txg&TXG_MASK], ==, 0);
+ ASSERT3U(dn->dn_next_bonustype[txg&TXG_MASK], ==, 0);
dprintf_ds(os->os_dsl_dataset, "obj=%llu txg=%llu\n",
dn->dn_object, txg);
for (db = list_head(&dn->dn_dbufs); db; db = db_next) {
db_next = list_next(&dn->dn_dbufs, db);
- if (db->db_blkid != 0 && db->db_blkid != DB_BONUS_BLKID) {
+ if (db->db_blkid != 0 && db->db_blkid != DMU_BONUS_BLKID &&
+ db->db_blkid != DMU_SPILL_BLKID) {
mutex_exit(&dn->dn_dbufs_mtx);
goto fail;
}
int epbs, new_nlevels;
uint64_t sz;
- ASSERT(blkid != DB_BONUS_BLKID);
+ ASSERT(blkid != DMU_BONUS_BLKID);
ASSERT(have_read ?
RW_READ_HELD(&dn->dn_struct_rwlock) :
/* dirty the left indirects */
db = dbuf_hold_level(dn, old_nlevels, 0, FTAG);
+ ASSERT(db != NULL);
new = dbuf_dirty(db, tx);
dbuf_rele(db, FTAG);
for (dr = list_head(list); dr; dr = dr_next) {
dr_next = list_next(&dn->dn_dirty_records[txgoff], dr);
if (dr->dr_dbuf->db_level != new_nlevels-1 &&
- dr->dr_dbuf->db_blkid != DB_BONUS_BLKID) {
+ dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID &&
+ dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) {
ASSERT(dr->dr_dbuf->db_level == old_nlevels-1);
list_remove(&dn->dn_dirty_records[txgoff], dr);
list_insert_tail(&new->dt.di.dr_children, dr);
rw_exit(&dn->dn_struct_rwlock);
}
+static boolean_t
+dnode_spill_freed(dnode_t *dn)
+{
+ int i;
+
+ mutex_enter(&dn->dn_mtx);
+ for (i = 0; i < TXG_SIZE; i++) {
+ if (dn->dn_rm_spillblk[i] == DN_KILL_SPILLBLK)
+ break;
+ }
+ mutex_exit(&dn->dn_mtx);
+ return (i < TXG_SIZE);
+}
+
/* return TRUE if this blkid was freed in a recent txg, or FALSE if it wasn't */
uint64_t
dnode_block_freed(dnode_t *dn, uint64_t blkid)
void *dp = spa_get_dsl(dn->dn_objset->os_spa);
int i;
- if (blkid == DB_BONUS_BLKID)
+ if (blkid == DMU_BONUS_BLKID)
return (FALSE);
/*
if (dn->dn_free_txg)
return (TRUE);
- /*
- * If dn_datablkshift is not set, then there's only a single
- * block, in which case there will never be a free range so it
- * won't matter.
- */
+ if (blkid == DMU_SPILL_BLKID)
+ return (dnode_spill_freed(dn));
+
range_tofind.fr_blkid = blkid;
mutex_enter(&dn->dn_mtx);
for (i = 0; i < TXG_SIZE; i++) {
void
dnode_willuse_space(dnode_t *dn, int64_t space, dmu_tx_t *tx)
{
- objset_impl_t *os = dn->dn_objset;
+ objset_t *os = dn->dn_objset;
dsl_dataset_t *ds = os->os_dsl_dataset;
if (space > 0)
dmu_tx_willuse_space(tx, space);
}
+/*
+ * This function scans a block at the indicated "level" looking for
+ * a hole or data (depending on 'flags'). If level > 0, then we are
+ * scanning an indirect block looking at its pointers. If level == 0,
+ * then we are looking at a block of dnodes. If we don't find what we
+ * are looking for in the block, we return ESRCH. Otherwise, return
+ * with *offset pointing to the beginning (if searching forwards) or
+ * end (if searching backwards) of the range covered by the block
+ * pointer we matched on (or dnode).
+ *
+ * The basic search algorithm used below by dnode_next_offset() is to
+ * use this function to search up the block tree (widen the search) until
+ * we find something (i.e., we don't return ESRCH) and then search back
+ * down the tree (narrow the search) until we reach our original search
+ * level.
+ */
static int
dnode_next_offset_level(dnode_t *dn, int flags, uint64_t *offset,
int lvl, uint64_t blkfill, uint64_t txg)
dprintf("probing object %llu offset %llx level %d of %u\n",
dn->dn_object, *offset, lvl, dn->dn_phys->dn_nlevels);
- hole = flags & DNODE_FIND_HOLE;
+ hole = ((flags & DNODE_FIND_HOLE) != 0);
inc = (flags & DNODE_FIND_BACKWARDS) ? -1 : 1;
ASSERT(txg == 0 || !hole);
for (i = (*offset >> span) & (blkfill - 1);
i >= 0 && i < blkfill; i += inc) {
- boolean_t newcontents = B_TRUE;
- if (txg) {
- int j;
- newcontents = B_FALSE;
- for (j = 0; j < dnp[i].dn_nblkptr; j++) {
- if (dnp[i].dn_blkptr[j].blk_birth > txg)
- newcontents = B_TRUE;
- }
- }
- if (!dnp[i].dn_type == hole && newcontents)
+ if ((dnp[i].dn_type == DMU_OT_NONE) == hole)
break;
*offset += (1ULL << span) * inc;
}
error = ESRCH;
} else {
blkptr_t *bp = data;
+ uint64_t start = *offset;
span = (lvl - 1) * epbs + dn->dn_datablkshift;
minfill = 0;
maxfill = blkfill << ((lvl - 1) * epbs);
else
minfill++;
- for (i = (*offset >> span) & ((1ULL << epbs) - 1);
+ *offset = *offset >> span;
+ for (i = BF64_GET(*offset, 0, epbs);
i >= 0 && i < epb; i += inc) {
if (bp[i].blk_fill >= minfill &&
bp[i].blk_fill <= maxfill &&
(hole || bp[i].blk_birth > txg))
break;
- if (inc < 0 && *offset < (1ULL << span))
- *offset = 0;
- else
- *offset += (1ULL << span) * inc;
+ if (inc > 0 || *offset > 0)
+ *offset += inc;
+ }
+ *offset = *offset << span;
+ if (inc < 0) {
+ /* traversing backwards; position offset at the end */
+ ASSERT3U(*offset, <=, start);
+ *offset = MIN(*offset + (1ULL << span) - 1, start);
+ } else if (*offset < start) {
+ *offset = start;
}
- if (i < 0 || i == epb)
+ if (i < 0 || i >= epb)
error = ESRCH;
}