{ "dnode_hold_free_lock_retry", KSTAT_DATA_UINT64 },
{ "dnode_hold_free_overflow", KSTAT_DATA_UINT64 },
{ "dnode_hold_free_refcount", KSTAT_DATA_UINT64 },
- { "dnode_hold_free_txg", KSTAT_DATA_UINT64 },
{ "dnode_free_interior_lock_retry", KSTAT_DATA_UINT64 },
{ "dnode_allocate", KSTAT_DATA_UINT64 },
{ "dnode_reallocate", KSTAT_DATA_UINT64 },
* Every dbuf has a reference, and dropping a tracked reference is
* O(number of references), so don't track dn_holds.
*/
- refcount_create_untracked(&dn->dn_holds);
- refcount_create(&dn->dn_tx_holds);
+ zfs_refcount_create_untracked(&dn->dn_holds);
+ zfs_refcount_create(&dn->dn_tx_holds);
list_link_init(&dn->dn_link);
bzero(&dn->dn_next_nblkptr[0], sizeof (dn->dn_next_nblkptr));
mutex_destroy(&dn->dn_mtx);
mutex_destroy(&dn->dn_dbufs_mtx);
cv_destroy(&dn->dn_notxholds);
- refcount_destroy(&dn->dn_holds);
- refcount_destroy(&dn->dn_tx_holds);
+ zfs_refcount_destroy(&dn->dn_holds);
+ zfs_refcount_destroy(&dn->dn_tx_holds);
ASSERT(!list_link_active(&dn->dn_link));
for (i = 0; i < TXG_SIZE; i++) {
void
dnode_setbonuslen(dnode_t *dn, int newsize, dmu_tx_t *tx)
{
- ASSERT3U(refcount_count(&dn->dn_holds), >=, 1);
+ ASSERT3U(zfs_refcount_count(&dn->dn_holds), >=, 1);
dnode_setdirty(dn, tx);
rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
ASSERT3U(newsize, <=, DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) -
(dn->dn_nblkptr-1) * sizeof (blkptr_t));
+
+ if (newsize < dn->dn_bonuslen) {
+ /* clear any data after the end of the new size */
+ size_t diff = dn->dn_bonuslen - newsize;
+ char *data_end = ((char *)dn->dn_bonus->db.db_data) + newsize;
+ bzero(data_end, diff);
+ }
+
dn->dn_bonuslen = newsize;
if (newsize == 0)
dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = DN_ZERO_BONUSLEN;
void
dnode_setbonus_type(dnode_t *dn, dmu_object_type_t newtype, dmu_tx_t *tx)
{
- ASSERT3U(refcount_count(&dn->dn_holds), >=, 1);
+ ASSERT3U(zfs_refcount_count(&dn->dn_holds), >=, 1);
dnode_setdirty(dn, tx);
rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
dn->dn_bonustype = newtype;
void
dnode_rm_spill(dnode_t *dn, dmu_tx_t *tx)
{
- ASSERT3U(refcount_count(&dn->dn_holds), >=, 1);
+ ASSERT3U(zfs_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_rm_spillblk[tx->tx_txg & TXG_MASK] = DN_KILL_SPILLBLK;
dn->dn_have_spill = B_FALSE;
}
ASSERT0(dn->dn_allocated_txg);
ASSERT0(dn->dn_assigned_txg);
ASSERT0(dn->dn_dirty_txg);
- ASSERT(refcount_is_zero(&dn->dn_tx_holds));
- ASSERT3U(refcount_count(&dn->dn_holds), <=, 1);
+ ASSERT(zfs_refcount_is_zero(&dn->dn_tx_holds));
+ ASSERT3U(zfs_refcount_count(&dn->dn_holds), <=, 1);
ASSERT(avl_is_empty(&dn->dn_dbufs));
for (i = 0; i < TXG_SIZE; i++) {
void
dnode_reallocate(dnode_t *dn, dmu_object_type_t ot, int blocksize,
- dmu_object_type_t bonustype, int bonuslen, int dn_slots, dmu_tx_t *tx)
+ dmu_object_type_t bonustype, int bonuslen, int dn_slots,
+ boolean_t keep_spill, dmu_tx_t *tx)
{
int nblkptr;
dnode_setdirty(dn, tx);
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)));
+ ASSERT0(dn->dn_maxblkid);
+ ASSERT(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;
+ 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;
+ dn->dn_next_bonuslen[tx->tx_txg & TXG_MASK] = bonuslen;
if (bonustype == DMU_OT_SA) /* Maximize bonus space for SA */
nblkptr = 1;
1 + ((DN_SLOTS_TO_BONUSLEN(dn_slots) - bonuslen) >>
SPA_BLKPTRSHIFT));
if (dn->dn_bonustype != bonustype)
- dn->dn_next_bonustype[tx->tx_txg&TXG_MASK] = 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) {
+ dn->dn_next_nblkptr[tx->tx_txg & TXG_MASK] = nblkptr;
+ if (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR && !keep_spill) {
dbuf_rm_spill(dn, tx);
dnode_rm_spill(dn, tx);
}
+
rw_exit(&dn->dn_struct_rwlock);
/* change type */
ASSERT(!RW_LOCK_HELD(&odn->dn_struct_rwlock));
ASSERT(MUTEX_NOT_HELD(&odn->dn_mtx));
ASSERT(MUTEX_NOT_HELD(&odn->dn_dbufs_mtx));
- ASSERT(!RW_LOCK_HELD(&odn->dn_zfetch.zf_rwlock));
+ ASSERT(!MUTEX_HELD(&odn->dn_zfetch.zf_lock));
/* Copy fields. */
ndn->dn_objset = odn->dn_objset;
ndn->dn_dirty_txg = odn->dn_dirty_txg;
ndn->dn_dirtyctx = odn->dn_dirtyctx;
ndn->dn_dirtyctx_firstset = odn->dn_dirtyctx_firstset;
- ASSERT(refcount_count(&odn->dn_tx_holds) == 0);
- refcount_transfer(&ndn->dn_holds, &odn->dn_holds);
+ ASSERT(zfs_refcount_count(&odn->dn_tx_holds) == 0);
+ zfs_refcount_transfer(&ndn->dn_holds, &odn->dn_holds);
ASSERT(avl_is_empty(&ndn->dn_dbufs));
avl_swap(&ndn->dn_dbufs, &odn->dn_dbufs);
ndn->dn_dbufs_count = odn->dn_dbufs_count;
offsetof(dmu_buf_impl_t, db_link));
odn->dn_dbufs_count = 0;
odn->dn_bonus = NULL;
- odn->dn_zfetch.zf_dnode = NULL;
+ dmu_zfetch_fini(&odn->dn_zfetch);
/*
* Set the low bit of the objset pointer to ensure that dnode_move()
* hold before the dbuf is removed, the hold is discounted, and the
* removal is blocked until the move completes.
*/
- refcount = refcount_count(&odn->dn_holds);
+ refcount = zfs_refcount_count(&odn->dn_holds);
ASSERT(refcount >= 0);
dbufs = odn->dn_dbufs_count;
list_link_replace(&odn->dn_link, &ndn->dn_link);
/* If the dnode was safe to move, the refcount cannot have changed. */
- ASSERT(refcount == refcount_count(&ndn->dn_holds));
+ ASSERT(refcount == zfs_refcount_count(&ndn->dn_holds));
ASSERT(dbufs == ndn->dn_dbufs_count);
zrl_exit(&ndn->dn_handle->dnh_zrlock); /* handle has moved */
mutex_exit(&os->os_lock);
} else if (DN_SLOT_IS_PTR(dn)) {
mutex_enter(&dn->dn_mtx);
boolean_t can_free = (dn->dn_type == DMU_OT_NONE &&
+ zfs_refcount_is_zero(&dn->dn_holds) &&
!DNODE_IS_DIRTY(dn));
mutex_exit(&dn->dn_mtx);
ASSERT3S(idx + slots, <=, DNODES_PER_BLOCK);
- while (!dnode_slots_tryenter(children, idx, slots))
+ while (!dnode_slots_tryenter(children, idx, slots)) {
DNODE_STAT_BUMP(dnode_free_interior_lock_retry);
+ cond_resched();
+ }
dnode_set_slots(children, idx, slots, DN_SLOT_FREE);
dnode_slots_rele(children, idx, slots);
* has a hold on this dnode while we are trying to evict this
* dnode.
*/
- while (refcount_count(&dn->dn_holds) > 0)
+ while (zfs_refcount_count(&dn->dn_holds) > 0)
delay(1);
ASSERT(dn->dn_dbuf == NULL ||
dmu_buf_get_user(&dn->dn_dbuf->db) == NULL);
* it wouldn't be eligible for eviction and this function
* would not have been called.
*/
- ASSERT(refcount_is_zero(&dn->dn_holds));
- ASSERT(refcount_is_zero(&dn->dn_tx_holds));
+ ASSERT(zfs_refcount_is_zero(&dn->dn_holds));
+ ASSERT(zfs_refcount_is_zero(&dn->dn_tx_holds));
dnode_destroy(dn); /* implicit zrl_remove() for first slot */
zrl_destroy(&dnh->dnh_zrlock);
* as an extra dnode slot by an large dnode, in which case it returns
* ENOENT.
*
+ * If the DNODE_DRY_RUN flag is set, we don't actually hold the dnode, just
+ * return whether the hold would succeed or not. tag and dnp should set to
+ * NULL in this case.
+ *
* errors:
* EINVAL - Invalid object number or flags.
* ENOSPC - Hole too small to fulfill "slots" request (DNODE_MUST_BE_FREE)
* EEXIST - Refers to an allocated dnode (DNODE_MUST_BE_FREE)
+ * - Refers to a freeing dnode (DNODE_MUST_BE_FREE)
* - Refers to an interior dnode slot (DNODE_MUST_BE_ALLOCATED)
* ENOENT - The requested dnode is not allocated (DNODE_MUST_BE_ALLOCATED)
+ * - The requested dnode is being freed (DNODE_MUST_BE_ALLOCATED)
* EIO - I/O error when reading the meta dnode dbuf.
*
* succeeds even for free dnodes.
ASSERT(!(flag & DNODE_MUST_BE_ALLOCATED) || (slots == 0));
ASSERT(!(flag & DNODE_MUST_BE_FREE) || (slots > 0));
+ IMPLY(flag & DNODE_DRY_RUN, (tag == NULL) && (dnp == NULL));
/*
* If you are holding the spa config lock as writer, you shouldn't
if ((flag & DNODE_MUST_BE_FREE) && type != DMU_OT_NONE)
return (SET_ERROR(EEXIST));
DNODE_VERIFY(dn);
- (void) refcount_add(&dn->dn_holds, tag);
- *dnp = dn;
+ /* Don't actually hold if dry run, just return 0 */
+ if (!(flag & DNODE_DRY_RUN)) {
+ (void) zfs_refcount_add(&dn->dn_holds, tag);
+ *dnp = dn;
+ }
return (0);
}
}
blk = dbuf_whichblock(mdn, 0, object * sizeof (dnode_phys_t));
-
db = dbuf_hold(mdn, blk, FTAG);
if (drop_struct_lock)
rw_exit(&mdn->dn_struct_rwlock);
}
ASSERT(dnc->dnc_count == epb);
- dn = DN_SLOT_UNINIT;
if (flag & DNODE_MUST_BE_ALLOCATED) {
slots = 1;
- while (dn == DN_SLOT_UNINIT) {
- dnode_slots_hold(dnc, idx, slots);
- dnh = &dnc->dnc_children[idx];
-
- if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) {
- dn = dnh->dnh_dnode;
- break;
- } else if (dnh->dnh_dnode == DN_SLOT_INTERIOR) {
- DNODE_STAT_BUMP(dnode_hold_alloc_interior);
- dnode_slots_rele(dnc, idx, slots);
- dbuf_rele(db, FTAG);
- return (SET_ERROR(EEXIST));
- } else if (dnh->dnh_dnode != DN_SLOT_ALLOCATED) {
- DNODE_STAT_BUMP(dnode_hold_alloc_misses);
- dnode_slots_rele(dnc, idx, slots);
- dbuf_rele(db, FTAG);
- return (SET_ERROR(ENOENT));
- }
+ dnode_slots_hold(dnc, idx, slots);
+ dnh = &dnc->dnc_children[idx];
+ if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) {
+ dn = dnh->dnh_dnode;
+ } else if (dnh->dnh_dnode == DN_SLOT_INTERIOR) {
+ DNODE_STAT_BUMP(dnode_hold_alloc_interior);
+ dnode_slots_rele(dnc, idx, slots);
+ dbuf_rele(db, FTAG);
+ return (SET_ERROR(EEXIST));
+ } else if (dnh->dnh_dnode != DN_SLOT_ALLOCATED) {
+ DNODE_STAT_BUMP(dnode_hold_alloc_misses);
+ dnode_slots_rele(dnc, idx, slots);
+ dbuf_rele(db, FTAG);
+ return (SET_ERROR(ENOENT));
+ } else {
dnode_slots_rele(dnc, idx, slots);
- if (!dnode_slots_tryenter(dnc, idx, slots)) {
+ while (!dnode_slots_tryenter(dnc, idx, slots)) {
DNODE_STAT_BUMP(dnode_hold_alloc_lock_retry);
- continue;
+ cond_resched();
}
/*
}
mutex_enter(&dn->dn_mtx);
- if (dn->dn_type == DMU_OT_NONE) {
+ if (dn->dn_type == DMU_OT_NONE || dn->dn_free_txg != 0) {
DNODE_STAT_BUMP(dnode_hold_alloc_type_none);
mutex_exit(&dn->dn_mtx);
dnode_slots_rele(dnc, idx, slots);
return (SET_ERROR(ENOENT));
}
+ /* Don't actually hold if dry run, just return 0 */
+ if (flag & DNODE_DRY_RUN) {
+ mutex_exit(&dn->dn_mtx);
+ dnode_slots_rele(dnc, idx, slots);
+ dbuf_rele(db, FTAG);
+ return (0);
+ }
+
DNODE_STAT_BUMP(dnode_hold_alloc_hits);
} else if (flag & DNODE_MUST_BE_FREE) {
return (SET_ERROR(ENOSPC));
}
- while (dn == DN_SLOT_UNINIT) {
- dnode_slots_hold(dnc, idx, slots);
-
- if (!dnode_check_slots_free(dnc, idx, slots)) {
- DNODE_STAT_BUMP(dnode_hold_free_misses);
- dnode_slots_rele(dnc, idx, slots);
- dbuf_rele(db, FTAG);
- return (SET_ERROR(ENOSPC));
- }
+ dnode_slots_hold(dnc, idx, slots);
+ if (!dnode_check_slots_free(dnc, idx, slots)) {
+ DNODE_STAT_BUMP(dnode_hold_free_misses);
dnode_slots_rele(dnc, idx, slots);
- if (!dnode_slots_tryenter(dnc, idx, slots)) {
- DNODE_STAT_BUMP(dnode_hold_free_lock_retry);
- continue;
- }
+ dbuf_rele(db, FTAG);
+ return (SET_ERROR(ENOSPC));
+ }
- if (!dnode_check_slots_free(dnc, idx, slots)) {
- DNODE_STAT_BUMP(dnode_hold_free_lock_misses);
- dnode_slots_rele(dnc, idx, slots);
- dbuf_rele(db, FTAG);
- return (SET_ERROR(ENOSPC));
- }
+ dnode_slots_rele(dnc, idx, slots);
+ while (!dnode_slots_tryenter(dnc, idx, slots)) {
+ DNODE_STAT_BUMP(dnode_hold_free_lock_retry);
+ cond_resched();
+ }
- /*
- * Allocated but otherwise free dnodes which would
- * be in the interior of a multi-slot dnodes need
- * to be freed. Single slot dnodes can be safely
- * re-purposed as a performance optimization.
- */
- if (slots > 1)
- dnode_reclaim_slots(dnc, idx + 1, slots - 1);
+ if (!dnode_check_slots_free(dnc, idx, slots)) {
+ DNODE_STAT_BUMP(dnode_hold_free_lock_misses);
+ dnode_slots_rele(dnc, idx, slots);
+ dbuf_rele(db, FTAG);
+ return (SET_ERROR(ENOSPC));
+ }
- dnh = &dnc->dnc_children[idx];
- if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) {
- dn = dnh->dnh_dnode;
- } else {
- dn = dnode_create(os, dn_block + idx, db,
- object, dnh);
- }
+ /*
+ * Allocated but otherwise free dnodes which would
+ * be in the interior of a multi-slot dnodes need
+ * to be freed. Single slot dnodes can be safely
+ * re-purposed as a performance optimization.
+ */
+ if (slots > 1)
+ dnode_reclaim_slots(dnc, idx + 1, slots - 1);
+
+ dnh = &dnc->dnc_children[idx];
+ if (DN_SLOT_IS_PTR(dnh->dnh_dnode)) {
+ dn = dnh->dnh_dnode;
+ } else {
+ dn = dnode_create(os, dn_block + idx, db,
+ object, dnh);
}
mutex_enter(&dn->dn_mtx);
- if (!refcount_is_zero(&dn->dn_holds)) {
+ if (!zfs_refcount_is_zero(&dn->dn_holds) || dn->dn_free_txg) {
DNODE_STAT_BUMP(dnode_hold_free_refcount);
mutex_exit(&dn->dn_mtx);
dnode_slots_rele(dnc, idx, slots);
return (SET_ERROR(EEXIST));
}
+ /* Don't actually hold if dry run, just return 0 */
+ if (flag & DNODE_DRY_RUN) {
+ mutex_exit(&dn->dn_mtx);
+ dnode_slots_rele(dnc, idx, slots);
+ dbuf_rele(db, FTAG);
+ return (0);
+ }
+
dnode_set_slots(dnc, idx + 1, slots - 1, DN_SLOT_INTERIOR);
DNODE_STAT_BUMP(dnode_hold_free_hits);
} else {
return (SET_ERROR(EINVAL));
}
- if (dn->dn_free_txg) {
- DNODE_STAT_BUMP(dnode_hold_free_txg);
- type = dn->dn_type;
- mutex_exit(&dn->dn_mtx);
- dnode_slots_rele(dnc, idx, slots);
- dbuf_rele(db, FTAG);
- return (SET_ERROR((flag & DNODE_MUST_BE_ALLOCATED) ?
- ENOENT : EEXIST));
- }
+ ASSERT0(dn->dn_free_txg);
- if (refcount_add(&dn->dn_holds, tag) == 1)
+ if (zfs_refcount_add(&dn->dn_holds, tag) == 1)
dbuf_add_ref(db, dnh);
mutex_exit(&dn->dn_mtx);
dnode_add_ref(dnode_t *dn, void *tag)
{
mutex_enter(&dn->dn_mtx);
- if (refcount_is_zero(&dn->dn_holds)) {
+ if (zfs_refcount_is_zero(&dn->dn_holds)) {
mutex_exit(&dn->dn_mtx);
return (FALSE);
}
- VERIFY(1 < refcount_add(&dn->dn_holds, tag));
+ VERIFY(1 < zfs_refcount_add(&dn->dn_holds, tag));
mutex_exit(&dn->dn_mtx);
return (TRUE);
}
dmu_buf_impl_t *db = dn->dn_dbuf;
dnode_handle_t *dnh = dn->dn_handle;
- refs = refcount_remove(&dn->dn_holds, tag);
+ refs = zfs_refcount_remove(&dn->dn_holds, tag);
mutex_exit(&dn->dn_mtx);
/*
}
}
+/*
+ * Test whether we can create a dnode at the specified location.
+ */
+int
+dnode_try_claim(objset_t *os, uint64_t object, int slots)
+{
+ return (dnode_hold_impl(os, object, DNODE_MUST_BE_FREE | DNODE_DRY_RUN,
+ slots, NULL, NULL));
+}
+
void
dnode_setdirty(dnode_t *dn, dmu_tx_t *tx)
{
return;
}
- ASSERT(!refcount_is_zero(&dn->dn_holds) ||
+ ASSERT(!zfs_refcount_is_zero(&dn->dn_holds) ||
!avl_is_empty(&dn->dn_dbufs));
ASSERT(dn->dn_datablksz != 0);
- ASSERT0(dn->dn_next_bonuslen[txg&TXG_MASK]);
- ASSERT0(dn->dn_next_blksz[txg&TXG_MASK]);
- ASSERT0(dn->dn_next_bonustype[txg&TXG_MASK]);
+ ASSERT0(dn->dn_next_bonuslen[txg & TXG_MASK]);
+ ASSERT0(dn->dn_next_blksz[txg & TXG_MASK]);
+ ASSERT0(dn->dn_next_bonustype[txg & TXG_MASK]);
dprintf_ds(os->os_dsl_dataset, "obj=%llu txg=%llu\n",
dn->dn_object, txg);
/* resize the old block */
err = dbuf_hold_impl(dn, 0, 0, TRUE, FALSE, FTAG, &db);
- if (err == 0)
+ if (err == 0) {
dbuf_new_size(db, size, tx);
- else if (err != ENOENT)
+ } else if (err != ENOENT) {
goto fail;
+ }
dnode_setdblksz(dn, size);
dnode_setdirty(dn, tx);
dn->dn_indblkshift = ibs;
dn->dn_next_indblkshift[tx->tx_txg&TXG_MASK] = ibs;
}
- /* rele after we have fixed the blocksize in the dnode */
+ /* release after we have fixed the blocksize in the dnode */
if (db)
dbuf_rele(db, FTAG);
/* read-holding callers must not rely on the lock being continuously held */
void
-dnode_new_blkid(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx, boolean_t have_read)
+dnode_new_blkid(dnode_t *dn, uint64_t blkid, dmu_tx_t *tx, boolean_t have_read,
+ boolean_t force)
{
int epbs, new_nlevels;
uint64_t sz;
}
}
- if (blkid <= dn->dn_maxblkid)
+ /*
+ * Raw sends (indicated by the force flag) require that we take the
+ * given blkid even if the value is lower than the current value.
+ */
+ if (!force && blkid <= dn->dn_maxblkid)
goto out;
+ /*
+ * We use the (otherwise unused) top bit of dn_next_maxblkid[txgoff]
+ * to indicate that this field is set. This allows us to set the
+ * maxblkid to 0 on an existing object in dnode_sync().
+ */
dn->dn_maxblkid = blkid;
- dn->dn_next_maxblkid[tx->tx_txg & TXG_MASK] = blkid;
+ dn->dn_next_maxblkid[tx->tx_txg & TXG_MASK] =
+ blkid | DMU_NEXT_MAXBLKID_SET;
/*
* Compute the number of levels necessary to support the new maxblkid.
+ * Raw sends will ensure nlevels is set correctly for us.
*/
new_nlevels = 1;
epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
ASSERT3U(new_nlevels, <=, DN_MAX_LEVELS);
- if (new_nlevels > dn->dn_nlevels)
- dnode_set_nlevels_impl(dn, new_nlevels, tx);
+ if (!force) {
+ if (new_nlevels > dn->dn_nlevels)
+ dnode_set_nlevels_impl(dn, new_nlevels, tx);
+ } else {
+ ASSERT3U(dn->dn_nlevels, >=, new_nlevels);
+ }
out:
if (have_read)
for (; db != NULL; db = AVL_NEXT(&dn->dn_dbufs, db)) {
if (db->db_level != 1 || db->db_blkid >= end_blkid)
break;
- ASSERT(db->db_dirtycnt > 0);
+ if (db->db_state != DB_EVICTING)
+ ASSERT(db->db_dirtycnt > 0);
}
#endif
mutex_exit(&dn->dn_dbufs_mtx);
int trunc = FALSE;
int epbs;
- rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
blksz = dn->dn_datablksz;
blkshift = dn->dn_datablkshift;
epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
head = P2NPHASE(off, blksz);
blkoff = P2PHASE(off, blksz);
if ((off >> blkshift) > dn->dn_maxblkid)
- goto out;
+ return;
} else {
ASSERT(dn->dn_maxblkid == 0);
if (off == 0 && len >= blksz) {
*/
blkid = 0;
nblks = 1;
- if (dn->dn_nlevels > 1)
+ if (dn->dn_nlevels > 1) {
+ rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
dnode_dirty_l1(dn, 0, tx);
+ rw_exit(&dn->dn_struct_rwlock);
+ }
goto done;
} else if (off >= blksz) {
/* Freeing past end-of-data */
- goto out;
+ return;
} else {
/* Freeing part of the block. */
head = blksz - off;
}
/* zero out any partial block data at the start of the range */
if (head) {
+ int res;
ASSERT3U(blkoff + head, ==, blksz);
if (len < head)
head = len;
- if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off),
- TRUE, FALSE, FTAG, &db) == 0) {
+ rw_enter(&dn->dn_struct_rwlock, RW_READER);
+ res = dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off),
+ TRUE, FALSE, FTAG, &db);
+ rw_exit(&dn->dn_struct_rwlock);
+ if (res == 0) {
caddr_t data;
+ boolean_t dirty;
+ db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_READER,
+ FTAG);
/* don't dirty if it isn't on disk and isn't dirty */
- if (db->db_last_dirty ||
- (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) {
- rw_exit(&dn->dn_struct_rwlock);
+ dirty = db->db_last_dirty ||
+ (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr));
+ dmu_buf_unlock_parent(db, dblt, FTAG);
+ if (dirty) {
dmu_buf_will_dirty(&db->db, tx);
- rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
data = db->db.db_data;
bzero(data + blkoff, head);
}
/* If the range was less than one block, we're done */
if (len == 0)
- goto out;
+ return;
/* If the remaining range is past end of file, we're done */
if ((off >> blkshift) > dn->dn_maxblkid)
- goto out;
+ return;
ASSERT(ISP2(blksz));
if (trunc)
ASSERT0(P2PHASE(off, blksz));
/* zero out any partial block data at the end of the range */
if (tail) {
+ int res;
if (len < tail)
tail = len;
- if (dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off+len),
- TRUE, FALSE, FTAG, &db) == 0) {
+ rw_enter(&dn->dn_struct_rwlock, RW_READER);
+ res = dbuf_hold_impl(dn, 0, dbuf_whichblock(dn, 0, off+len),
+ TRUE, FALSE, FTAG, &db);
+ rw_exit(&dn->dn_struct_rwlock);
+ if (res == 0) {
+ boolean_t dirty;
/* don't dirty if not on disk and not dirty */
- if (db->db_last_dirty ||
- (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr))) {
- rw_exit(&dn->dn_struct_rwlock);
+ db_lock_type_t type = dmu_buf_lock_parent(db, RW_READER,
+ FTAG);
+ dirty = db->db_last_dirty ||
+ (db->db_blkptr && !BP_IS_HOLE(db->db_blkptr));
+ dmu_buf_unlock_parent(db, type, FTAG);
+ if (dirty) {
dmu_buf_will_dirty(&db->db, tx);
- rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
bzero(db->db.db_data, tail);
}
dbuf_rele(db, FTAG);
/* If the range did not include a full block, we are done */
if (len == 0)
- goto out;
+ return;
ASSERT(IS_P2ALIGNED(off, blksz));
ASSERT(trunc || IS_P2ALIGNED(len, blksz));
* amount of space if we copy the freed BPs into deadlists.
*/
if (dn->dn_nlevels > 1) {
+ rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
uint64_t first, last;
first = blkid >> epbs;
dnode_dirty_l1(dn, i, tx);
}
+ rw_exit(&dn->dn_struct_rwlock);
}
done:
dbuf_free_range(dn, blkid, blkid + nblks - 1, tx);
dnode_setdirty(dn, tx);
-out:
-
- rw_exit(&dn->dn_struct_rwlock);
}
static boolean_t
boolean_t hole;
int i, inc, error, span;
+ ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
+
hole = ((flags & DNODE_FIND_HOLE) != 0);
inc = (flags & DNODE_FIND_BACKWARDS) ? -1 : 1;
ASSERT(txg == 0 || !hole);
return (error);
}
data = db->db.db_data;
+ rw_enter(&db->db_rwlock, RW_READER);
}
-
if (db != NULL && txg != 0 && (db->db_blkptr == NULL ||
db->db_blkptr->blk_birth <= txg ||
BP_IS_HOLE(db->db_blkptr))) {
error = SET_ERROR(ESRCH);
}
- if (db)
+ if (db != NULL) {
+ rw_exit(&db->db_rwlock);
dbuf_rele(db, FTAG);
+ }
return (error);
}
return (error);
}
+
+#if defined(_KERNEL)
+EXPORT_SYMBOL(dnode_hold);
+EXPORT_SYMBOL(dnode_rele);
+EXPORT_SYMBOL(dnode_set_nlevels);
+EXPORT_SYMBOL(dnode_set_blksz);
+EXPORT_SYMBOL(dnode_free_range);
+EXPORT_SYMBOL(dnode_evict_dbufs);
+EXPORT_SYMBOL(dnode_evict_bonus);
+#endif
projects / mirror_zfs.git / blobdiff