int
dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)bonus;
dnode_t *dn;
int err;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
err = dmu_spill_hold_by_dnode(dn, DB_RF_CANFAIL, tag, dbp);
DB_DNODE_EXIT(db);
return (err);
}
dmu_object_type_t
dmu_get_bonustype(dmu_buf_t *db_fake)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
dnode_t *dn;
dmu_object_type_t type;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
type = dn->dn_bonustype;
DB_DNODE_EXIT(db);
return (type);
}
int
dmu_set_bonustype(dmu_buf_t *db_fake, dmu_object_type_t type, dmu_tx_t *tx)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
dnode_t *dn;
int error;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
if (!DMU_OT_IS_VALID(type)) {
error = SET_ERROR(EINVAL);
} else if (dn->dn_bonus != db) {
error = SET_ERROR(EINVAL);
} else {
dnode_setbonus_type(dn, type, tx);
error = 0;
}
DB_DNODE_EXIT(db);
return (error);
}
int
dmu_set_bonus(dmu_buf_t *db_fake, int newsize, dmu_tx_t *tx)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
dnode_t *dn;
int error;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
if (dn->dn_bonus != db) {
error = SET_ERROR(EINVAL);
} else if (newsize < 0 || newsize > db_fake->db_size) {
error = SET_ERROR(EINVAL);
} else {
dnode_setbonuslen(dn, newsize, tx);
error = 0;
}
DB_DNODE_EXIT(db);
return (error);
}
/*
* Decide whether block should be l2cached. Returns true if block is a ddt
* metadata and ddt metadata is cacheable, or if block isn't a ddt metadata
*/
boolean_t
dbuf_ddt_is_l2cacheable(dmu_buf_impl_t *db)
{
dmu_object_type_t ot;
spa_t *spa = db->db_objset->os_spa;
spa_meta_placement_t *mp = &spa->spa_meta_policy;
uint64_t specflags;
boolean_t match;
if (!spa_has_special(spa))
return (B_TRUE);
specflags = spa_specialclass_flags(db->db_objset);
match = !!(SPECIAL_FLAG_DATAMETA & specflags);
DB_DNODE_ENTER(db);
ot = DB_DNODE(db)->dn_type;
DB_DNODE_EXIT(db);
if ((!DMU_OT_IS_DDT_META(ot)) || (!match))
return (B_TRUE);
return (mp->spa_ddt_to_special != META_PLACEMENT_ON);
}
/*
* Decide whether block should be l2cached. Returns true if block's metadata
* type is l2cacheable or block isn't a metadata one
*/
boolean_t
dbuf_meta_is_l2cacheable(dmu_buf_impl_t *db)
{
boolean_t is_metadata, is_to_special;
dmu_object_type_t ot = DMU_OT_NONE;
spa_t *spa = db->db_objset->os_spa;
DB_DNODE_ENTER(db);
ot = DB_DNODE(db)->dn_type;
DB_DNODE_EXIT(db);
is_metadata = dmu_ot[ot].ot_metadata;
if (!is_metadata)
return (B_TRUE);
is_to_special = spa_meta_to_special(spa, db->db_objset, ot);
if (!is_to_special)
return (B_TRUE);
return (spa_meta_is_dual(spa, db->db_objset->os_zpl_meta_to_special,
ot));
}
/*
* add/remove/replace a single attribute and then rewrite the entire set
* of attributes.
*/
static int
sa_modify_attrs(sa_handle_t *hdl, sa_attr_type_t newattr,
sa_data_op_t action, sa_data_locator_t *locator, void *datastart,
uint16_t buflen, dmu_tx_t *tx)
{
sa_os_t *sa = hdl->sa_os->os_sa;
dmu_buf_impl_t *db = (dmu_buf_impl_t *)hdl->sa_bonus;
dnode_t *dn;
sa_bulk_attr_t *attr_desc;
void *old_data[2];
int bonus_attr_count = 0;
int bonus_data_size, spill_data_size;
int spill_attr_count = 0;
int error;
uint16_t length;
int i, j, k, length_idx;
sa_hdr_phys_t *hdr;
sa_idx_tab_t *idx_tab;
int attr_count;
int count;
ASSERT(MUTEX_HELD(&hdl->sa_lock));
/* First make of copy of the old data */
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
if (dn->dn_bonuslen != 0) {
bonus_data_size = hdl->sa_bonus->db_size;
old_data[0] = kmem_alloc(bonus_data_size, KM_SLEEP);
bcopy(hdl->sa_bonus->db_data, old_data[0],
hdl->sa_bonus->db_size);
bonus_attr_count = hdl->sa_bonus_tab->sa_layout->lot_attr_count;
} else {
old_data[0] = NULL;
}
DB_DNODE_EXIT(db);
/* Bring spill buffer online if it isn't currently */
if ((error = sa_get_spill(hdl)) == 0) {
spill_data_size = hdl->sa_spill->db_size;
old_data[1] = kmem_alloc(spill_data_size, KM_SLEEP);
bcopy(hdl->sa_spill->db_data, old_data[1],
hdl->sa_spill->db_size);
spill_attr_count =
hdl->sa_spill_tab->sa_layout->lot_attr_count;
} else if (error && error != ENOENT) {
if (old_data[0])
kmem_free(old_data[0], bonus_data_size);
return (error);
} else {
old_data[1] = NULL;
}
/* build descriptor of all attributes */
attr_count = bonus_attr_count + spill_attr_count;
if (action == SA_ADD)
attr_count++;
else if (action == SA_REMOVE)
attr_count--;
attr_desc = kmem_zalloc(sizeof (sa_bulk_attr_t) * attr_count, KM_SLEEP);
/*
* loop through bonus and spill buffer if it exists, and
* build up new attr_descriptor to reset the attributes
*/
k = j = 0;
count = bonus_attr_count;
hdr = SA_GET_HDR(hdl, SA_BONUS);
idx_tab = SA_IDX_TAB_GET(hdl, SA_BONUS);
for (; k != 2; k++) {
/* iterate over each attribute in layout */
for (i = 0, length_idx = 0; i != count; i++) {
sa_attr_type_t attr;
attr = idx_tab->sa_layout->lot_attrs[i];
if (attr == newattr) {
if (action == SA_REMOVE) {
j++;
continue;
}
ASSERT(SA_REGISTERED_LEN(sa, attr) == 0);
ASSERT(action == SA_REPLACE);
SA_ADD_BULK_ATTR(attr_desc, j, attr,
locator, datastart, buflen);
} else {
length = SA_REGISTERED_LEN(sa, attr);
if (length == 0) {
length = hdr->sa_lengths[length_idx++];
}
SA_ADD_BULK_ATTR(attr_desc, j, attr,
NULL, (void *)
(TOC_OFF(idx_tab->sa_idx_tab[attr]) +
(uintptr_t)old_data[k]), length);
}
}
if (k == 0 && hdl->sa_spill) {
hdr = SA_GET_HDR(hdl, SA_SPILL);
idx_tab = SA_IDX_TAB_GET(hdl, SA_SPILL);
count = spill_attr_count;
} else {
break;
}
}
if (action == SA_ADD) {
length = SA_REGISTERED_LEN(sa, newattr);
if (length == 0) {
length = buflen;
}
SA_ADD_BULK_ATTR(attr_desc, j, newattr, locator,
datastart, buflen);
}
error = sa_build_layouts(hdl, attr_desc, attr_count, tx);
if (old_data[0])
kmem_free(old_data[0], bonus_data_size);
if (old_data[1])
kmem_free(old_data[1], spill_data_size);
kmem_free(attr_desc, sizeof (sa_bulk_attr_t) * attr_count);
return (error);
}
static void
dnode_increase_indirection(dnode_t *dn, dmu_tx_t *tx)
{
dmu_buf_impl_t *db;
int txgoff = tx->tx_txg & TXG_MASK;
int nblkptr = dn->dn_phys->dn_nblkptr;
int old_toplvl = dn->dn_phys->dn_nlevels - 1;
int new_level = dn->dn_next_nlevels[txgoff];
int i;
rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
/* this dnode can't be paged out because it's dirty */
ASSERT(dn->dn_phys->dn_type != DMU_OT_NONE);
ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock));
ASSERT(new_level > 1 && dn->dn_phys->dn_nlevels > 0);
db = dbuf_hold_level(dn, dn->dn_phys->dn_nlevels, 0, FTAG);
ASSERT(db != NULL);
dn->dn_phys->dn_nlevels = new_level;
dprintf("os=%p obj=%llu, increase to %d\n", dn->dn_objset,
dn->dn_object, dn->dn_phys->dn_nlevels);
/* transfer dnode's block pointers to new indirect block */
(void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED|DB_RF_HAVESTRUCT);
ASSERT(db->db.db_data);
ASSERT(arc_released(db->db_buf));
ASSERT3U(sizeof (blkptr_t) * nblkptr, <=, db->db.db_size);
bcopy(dn->dn_phys->dn_blkptr, db->db.db_data,
sizeof (blkptr_t) * nblkptr);
arc_buf_freeze(db->db_buf);
/* set dbuf's parent pointers to new indirect buf */
for (i = 0; i < nblkptr; i++) {
dmu_buf_impl_t *child =
dbuf_find(dn->dn_objset, dn->dn_object, old_toplvl, i);
if (child == NULL)
continue;
#ifdef DEBUG
DB_DNODE_ENTER(child);
ASSERT3P(DB_DNODE(child), ==, dn);
DB_DNODE_EXIT(child);
#endif /* DEBUG */
if (child->db_parent && child->db_parent != dn->dn_dbuf) {
ASSERT(child->db_parent->db_level == db->db_level);
ASSERT(child->db_blkptr !=
&dn->dn_phys->dn_blkptr[child->db_blkid]);
mutex_exit(&child->db_mtx);
continue;
}
ASSERT(child->db_parent == NULL ||
child->db_parent == dn->dn_dbuf);
child->db_parent = db;
dbuf_add_ref(db, child);
if (db->db.db_data)
child->db_blkptr = (blkptr_t *)db->db.db_data + i;
else
child->db_blkptr = NULL;
dprintf_dbuf_bp(child, child->db_blkptr,
"changed db_blkptr to new indirect %s", "");
mutex_exit(&child->db_mtx);
}
bzero(dn->dn_phys->dn_blkptr, sizeof (blkptr_t) * nblkptr);
dbuf_rele(db, FTAG);
rw_exit(&dn->dn_struct_rwlock);
}
static void
free_children(dmu_buf_impl_t *db, uint64_t blkid, uint64_t nblks,
dmu_tx_t *tx)
{
dnode_t *dn;
blkptr_t *bp;
dmu_buf_impl_t *subdb;
uint64_t start, end, dbstart, dbend, i;
int epbs, shift;
/*
* There is a small possibility that this block will not be cached:
* 1 - if level > 1 and there are no children with level <= 1
* 2 - if this block was evicted since we read it from
* dmu_tx_hold_free().
*/
if (db->db_state != DB_CACHED)
(void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED);
dbuf_release_bp(db);
bp = db->db.db_data;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
shift = (db->db_level - 1) * epbs;
dbstart = db->db_blkid << epbs;
start = blkid >> shift;
if (dbstart < start) {
bp += start - dbstart;
} else {
start = dbstart;
}
dbend = ((db->db_blkid + 1) << epbs) - 1;
end = (blkid + nblks - 1) >> shift;
if (dbend <= end)
end = dbend;
ASSERT3U(start, <=, end);
if (db->db_level == 1) {
FREE_VERIFY(db, start, end, tx);
free_blocks(dn, bp, end-start+1, tx);
} else {
for (i = start; i <= end; i++, bp++) {
if (BP_IS_HOLE(bp))
continue;
rw_enter(&dn->dn_struct_rwlock, RW_READER);
VERIFY0(dbuf_hold_impl(dn, db->db_level - 1,
i, TRUE, FALSE, FTAG, &subdb));
rw_exit(&dn->dn_struct_rwlock);
ASSERT3P(bp, ==, subdb->db_blkptr);
free_children(subdb, blkid, nblks, tx);
dbuf_rele(subdb, FTAG);
}
}
/* If this whole block is free, free ourself too. */
for (i = 0, bp = db->db.db_data; i < 1 << epbs; i++, bp++) {
if (!BP_IS_HOLE(bp))
break;
}
if (i == 1 << epbs) {
/* didn't find any non-holes */
bzero(db->db.db_data, db->db.db_size);
free_blocks(dn, db->db_blkptr, 1, tx);
} else {
/*
* Partial block free; must be marked dirty so that it
* will be written out.
*/
ASSERT(db->db_dirtycnt > 0);
}
DB_DNODE_EXIT(db);
arc_buf_freeze(db->db_buf);
}
static void
free_verify(dmu_buf_impl_t *db, uint64_t start, uint64_t end, dmu_tx_t *tx)
{
int off, num;
int i, err, epbs;
uint64_t txg = tx->tx_txg;
dnode_t *dn;
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
off = start - (db->db_blkid * 1<<epbs);
num = end - start + 1;
ASSERT3U(off, >=, 0);
ASSERT3U(num, >=, 0);
ASSERT3U(db->db_level, >, 0);
ASSERT3U(db->db.db_size, ==, 1 << dn->dn_phys->dn_indblkshift);
ASSERT3U(off+num, <=, db->db.db_size >> SPA_BLKPTRSHIFT);
ASSERT(db->db_blkptr != NULL);
for (i = off; i < off+num; i++) {
uint64_t *buf;
dmu_buf_impl_t *child;
dbuf_dirty_record_t *dr;
int j;
ASSERT(db->db_level == 1);
rw_enter(&dn->dn_struct_rwlock, RW_READER);
err = dbuf_hold_impl(dn, db->db_level-1,
(db->db_blkid << epbs) + i, TRUE, FALSE, FTAG, &child);
rw_exit(&dn->dn_struct_rwlock);
if (err == ENOENT)
continue;
ASSERT(err == 0);
ASSERT(child->db_level == 0);
dr = child->db_last_dirty;
while (dr && dr->dr_txg > txg)
dr = dr->dr_next;
ASSERT(dr == NULL || dr->dr_txg == txg);
/* data_old better be zeroed */
if (dr) {
buf = dr->dt.dl.dr_data->b_data;
for (j = 0; j < child->db.db_size >> 3; j++) {
if (buf[j] != 0) {
panic("freed data not zero: "
"child=%p i=%d off=%d num=%d\n",
(void *)child, i, off, num);
}
}
}
/*
* db_data better be zeroed unless it's dirty in a
* future txg.
*/
mutex_enter(&child->db_mtx);
buf = child->db.db_data;
if (buf != NULL && child->db_state != DB_FILL &&
child->db_last_dirty == NULL) {
for (j = 0; j < child->db.db_size >> 3; j++) {
if (buf[j] != 0) {
panic("freed data not zero: "
"child=%p i=%d off=%d num=%d\n",
(void *)child, i, off, num);
}
}
}
mutex_exit(&child->db_mtx);
dbuf_rele(child, FTAG);
}
DB_DNODE_EXIT(db);
}
