void
fzap_upgrade(zap_t *zap, dmu_tx_t *tx, zap_flags_t flags)
{
dmu_buf_t *db;
zap_leaf_t *l;
int i;
zap_phys_t *zp;
ASSERT(RW_WRITE_HELD(&zap->zap_rwlock));
zap->zap_ismicro = FALSE;
(void) dmu_buf_update_user(zap->zap_dbuf, zap, zap,
&zap->zap_f.zap_phys, zap_evict);
mutex_init(&zap->zap_f.zap_num_entries_mtx, 0, 0, 0);
zap->zap_f.zap_block_shift = highbit64(zap->zap_dbuf->db_size) - 1;
zp = zap->zap_f.zap_phys;
/*
* explicitly zero it since it might be coming from an
* initialized microzap
*/
bzero(zap->zap_dbuf->db_data, zap->zap_dbuf->db_size);
zp->zap_block_type = ZBT_HEADER;
zp->zap_magic = ZAP_MAGIC;
zp->zap_ptrtbl.zt_shift = ZAP_EMBEDDED_PTRTBL_SHIFT(zap);
zp->zap_freeblk = 2; /* block 1 will be the first leaf */
zp->zap_num_leafs = 1;
zp->zap_num_entries = 0;
zp->zap_salt = zap->zap_salt;
zp->zap_normflags = zap->zap_normflags;
zp->zap_flags = flags;
/* block 1 will be the first leaf */
for (i = 0; i < (1<<zp->zap_ptrtbl.zt_shift); i++)
ZAP_EMBEDDED_PTRTBL_ENT(zap, i) = 1;
/*
* set up block 1 - the first leaf
*/
VERIFY(0 == dmu_buf_hold(zap->zap_objset, zap->zap_object,
1<<FZAP_BLOCK_SHIFT(zap), FTAG, &db, DMU_READ_NO_PREFETCH));
dmu_buf_will_dirty(db, tx);
l = kmem_zalloc(sizeof (zap_leaf_t), KM_SLEEP);
l->l_dbuf = db;
l->l_phys = db->db_data;
zap_leaf_init(l, zp->zap_normflags != 0);
kmem_free(l, sizeof (zap_leaf_t));
dmu_buf_rele(db, FTAG);
}
/*
* Link zp into dl. Can only fail if zp has been unlinked.
*/
int
zfs_link_create(zfs_dirlock_t *dl, znode_t *zp, dmu_tx_t *tx, int flag)
{
znode_t *dzp = dl->dl_dzp;
vnode_t *vp = ZTOV(zp);
uint64_t value;
int zp_is_dir = (vp->v_type == VDIR);
int error;
dmu_buf_will_dirty(zp->z_dbuf, tx);
mutex_enter(&zp->z_lock);
if (!(flag & ZRENAMING)) {
if (zp->z_unlinked) { /* no new links to unlinked zp */
ASSERT(!(flag & (ZNEW | ZEXISTS)));
mutex_exit(&zp->z_lock);
return (ENOENT);
}
zp->z_phys->zp_links++;
}
zp->z_phys->zp_parent = dzp->z_id; /* dzp is now zp's parent */
if (!(flag & ZNEW))
zfs_time_stamper_locked(zp, STATE_CHANGED, tx);
mutex_exit(&zp->z_lock);
dmu_buf_will_dirty(dzp->z_dbuf, tx);
mutex_enter(&dzp->z_lock);
dzp->z_phys->zp_size++; /* one dirent added */
dzp->z_phys->zp_links += zp_is_dir; /* ".." link from zp */
zfs_time_stamper_locked(dzp, CONTENT_MODIFIED, tx);
mutex_exit(&dzp->z_lock);
value = zfs_dirent(zp);
error = zap_add(zp->z_zfsvfs->z_os, dzp->z_id, dl->dl_name,
8, 1, &value, tx);
ASSERT(error == 0);
dnlc_update(ZTOV(dzp), dl->dl_name, vp);
return (0);
}
static int
zap_tryupgradedir(zap_t *zap, dmu_tx_t *tx)
{
if (RW_WRITE_HELD(&zap->zap_rwlock))
return (1);
if (rw_tryupgrade(&zap->zap_rwlock)) {
dmu_buf_will_dirty(zap->zap_dbuf, tx);
return (1);
}
return (0);
}
static int
zap_table_store(zap_t *zap, zap_table_phys_t *tbl, uint64_t idx, uint64_t val,
dmu_tx_t *tx)
{
int err;
uint64_t blk, off;
int bs = FZAP_BLOCK_SHIFT(zap);
dmu_buf_t *db;
ASSERT(RW_LOCK_HELD(&zap->zap_rwlock));
ASSERT(tbl->zt_blk != 0);
dprintf("storing %llx at index %llx\n", val, idx);
blk = idx >> (bs-3);
off = idx & ((1<<(bs-3))-1);
err = dmu_buf_hold(zap->zap_objset, zap->zap_object,
(tbl->zt_blk + blk) << bs, FTAG, &db, DMU_READ_NO_PREFETCH);
if (err)
return (err);
dmu_buf_will_dirty(db, tx);
if (tbl->zt_nextblk != 0) {
uint64_t idx2 = idx * 2;
uint64_t blk2 = idx2 >> (bs-3);
uint64_t off2 = idx2 & ((1<<(bs-3))-1);
dmu_buf_t *db2;
err = dmu_buf_hold(zap->zap_objset, zap->zap_object,
(tbl->zt_nextblk + blk2) << bs, FTAG, &db2,
DMU_READ_NO_PREFETCH);
if (err) {
dmu_buf_rele(db, FTAG);
return (err);
}
dmu_buf_will_dirty(db2, tx);
((uint64_t *)db2->db_data)[off2] = val;
((uint64_t *)db2->db_data)[off2+1] = val;
dmu_buf_rele(db2, FTAG);
}
static int
zfs_vfs_sync(struct mount *mp, __unused int waitfor, __unused vfs_context_t context)
{
zfsvfs_t *zfsvfs = vfs_fsprivate(mp);
ZFS_ENTER(zfsvfs);
/*
* Mac OS X needs a file system modify time
*
* We use the mtime of the "com.apple.system.mtime"
* extended attribute, which is associated with the
* file system root directory.
*
* Here we sync any mtime changes to this attribute.
*/
if (zfsvfs->z_mtime_vp != NULL) {
timestruc_t mtime;
znode_t *zp;
top:
zp = VTOZ(zfsvfs->z_mtime_vp);
ZFS_TIME_DECODE(&mtime, zp->z_phys->zp_mtime);
if (zfsvfs->z_last_mtime_synced < mtime.tv_sec) {
dmu_tx_t *tx;
int error;
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_bonus(tx, zp->z_id);
error = dmu_tx_assign(tx, zfsvfs->z_assign);
if (error) {
if (error == ERESTART && zfsvfs->z_assign == TXG_NOWAIT) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
} else {
dmu_buf_will_dirty(zp->z_dbuf, tx);
dmu_tx_commit(tx);
zfsvfs->z_last_mtime_synced = mtime.tv_sec;
}
}
}
if (zfsvfs->z_log != NULL)
zil_commit(zfsvfs->z_log, UINT64_MAX, 0);
else
txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0);
ZFS_EXIT(zfsvfs);
return (0);
}
int
zfs_make_xattrdir(znode_t *zp, vattr_t *vap, vnode_t **xvpp, cred_t *cr)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
znode_t *xzp;
dmu_tx_t *tx;
int error;
zfs_fuid_info_t *fuidp = NULL;
*xvpp = NULL;
if (error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, 0, B_FALSE, cr))
return (error);
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_bonus(tx, zp->z_id);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
if (IS_EPHEMERAL(crgetuid(cr)) || IS_EPHEMERAL(crgetgid(cr))) {
if (zfsvfs->z_fuid_obj == 0) {
dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
FUID_SIZE_ESTIMATE(zfsvfs));
dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, FALSE, NULL);
} else {
dmu_tx_hold_bonus(tx, zfsvfs->z_fuid_obj);
dmu_tx_hold_write(tx, zfsvfs->z_fuid_obj, 0,
FUID_SIZE_ESTIMATE(zfsvfs));
}
}
error = dmu_tx_assign(tx, zfsvfs->z_assign);
if (error) {
if (error == ERESTART && zfsvfs->z_assign == TXG_NOWAIT)
dmu_tx_wait(tx);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, tx, cr, IS_XATTR, &xzp, 0, NULL, &fuidp);
ASSERT(xzp->z_phys->zp_parent == zp->z_id);
dmu_buf_will_dirty(zp->z_dbuf, tx);
zp->z_phys->zp_xattr = xzp->z_id;
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp,
xzp, "", NULL, fuidp, vap);
if (fuidp)
zfs_fuid_info_free(fuidp);
dmu_tx_commit(tx);
*xvpp = ZTOV(xzp);
return (0);
}
static void
dsl_dataset_user_hold_sync_one_impl(nvlist_t *tmpholds, dsl_dataset_t *ds,
const char *htag, minor_t minor, uint64_t now, dmu_tx_t *tx)
{
dsl_pool_t *dp = ds->ds_dir->dd_pool;
objset_t *mos = dp->dp_meta_objset;
uint64_t zapobj;
ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock));
if (ds->ds_phys->ds_userrefs_obj == 0) {
/*
* This is the first user hold for this dataset. Create
* the userrefs zap object.
*/
dmu_buf_will_dirty(ds->ds_dbuf, tx);
zapobj = ds->ds_phys->ds_userrefs_obj =
zap_create(mos, DMU_OT_USERREFS, DMU_OT_NONE, 0, tx);
} else {
zapobj = ds->ds_phys->ds_userrefs_obj;
}
ds->ds_userrefs++;
VERIFY0(zap_add(mos, zapobj, htag, 8, 1, &now, tx));
if (minor != 0) {
char name[MAXNAMELEN];
nvlist_t *tags;
VERIFY0(dsl_pool_user_hold(dp, ds->ds_object,
htag, now, tx));
(void) snprintf(name, sizeof (name), "%llx",
(u_longlong_t)ds->ds_object);
if (nvlist_lookup_nvlist(tmpholds, name, &tags) != 0) {
VERIFY0(nvlist_alloc(&tags, NV_UNIQUE_NAME,
KM_PUSHPAGE));
fnvlist_add_boolean(tags, htag);
fnvlist_add_nvlist(tmpholds, name, tags);
fnvlist_free(tags);
} else {
fnvlist_add_boolean(tags, htag);
}
}
spa_history_log_internal_ds(ds, "hold", tx,
"tag=%s temp=%d refs=%llu",
htag, minor != 0, ds->ds_userrefs);
}
int
dmu_objset_userspace_upgrade(objset_t *os)
{
uint64_t obj;
int err = 0;
if (dmu_objset_userspace_present(os))
return (0);
if (!dmu_objset_userused_enabled(os->os))
return (ENOTSUP);
if (dmu_objset_is_snapshot(os))
return (EINVAL);
/*
* We simply need to mark every object dirty, so that it will be
* synced out and now accounted. If this is called
* concurrently, or if we already did some work before crashing,
* that's fine, since we track each object's accounted state
* independently.
*/
for (obj = 0; err == 0; err = dmu_object_next(os, &obj, FALSE, 0)) {
dmu_tx_t *tx;
dmu_buf_t *db;
int objerr;
if (issig(JUSTLOOKING) && issig(FORREAL))
return (EINTR);
objerr = dmu_bonus_hold(os, obj, FTAG, &db);
if (objerr)
continue;
tx = dmu_tx_create(os);
dmu_tx_hold_bonus(tx, obj);
objerr = dmu_tx_assign(tx, TXG_WAIT);
if (objerr) {
dmu_tx_abort(tx);
continue;
}
dmu_buf_will_dirty(db, tx);
dmu_buf_rele(db, FTAG);
dmu_tx_commit(tx);
}
os->os->os_flags |= OBJSET_FLAG_USERACCOUNTING_COMPLETE;
txg_wait_synced(dmu_objset_pool(os), 0);
return (0);
}
void
bplist_vacate(bplist_t *bpl, dmu_tx_t *tx)
{
mutex_enter(&bpl->bpl_lock);
ASSERT3P(bpl->bpl_queue, ==, NULL);
VERIFY(0 == bplist_hold(bpl));
dmu_buf_will_dirty(bpl->bpl_dbuf, tx);
VERIFY(0 == dmu_free_range(bpl->bpl_mos,
bpl->bpl_object, 0, -1ULL, tx));
bpl->bpl_phys->bpl_entries = 0;
bpl->bpl_phys->bpl_bytes = 0;
if (bpl->bpl_havecomp) {
bpl->bpl_phys->bpl_comp = 0;
bpl->bpl_phys->bpl_uncomp = 0;
}
mutex_exit(&bpl->bpl_lock);
}
void
dsl_dir_sync(dsl_dir_t *dd, dmu_tx_t *tx)
{
ASSERT(dmu_tx_is_syncing(tx));
dmu_buf_will_dirty(dd->dd_dbuf, tx);
mutex_enter(&dd->dd_lock);
ASSERT3U(dd->dd_tempreserved[tx->tx_txg&TXG_MASK], ==, 0);
dprintf_dd(dd, "txg=%llu towrite=%lluK\n", tx->tx_txg,
dd->dd_space_towrite[tx->tx_txg&TXG_MASK] / 1024);
dd->dd_space_towrite[tx->tx_txg&TXG_MASK] = 0;
mutex_exit(&dd->dd_lock);
/* release the hold from dsl_dir_dirty */
dmu_buf_rele(dd->dd_dbuf, dd);
}
int
zfs_make_xattrdir(znode_t *zp, vattr_t *vap, vnode_t **xvpp, cred_t *cr)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
znode_t *xzp;
dmu_tx_t *tx;
uint64_t xoid;
int error;
*xvpp = NULL;
#ifndef __APPLE__
/* In Mac OS X access preflighting is done above the file system. */
if (error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, cr))
return (error);
#endif /*!__APPLE__*/
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_bonus(tx, zp->z_id);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
error = dmu_tx_assign(tx, zfsvfs->z_assign);
if (error) {
if (error == ERESTART && zfsvfs->z_assign == TXG_NOWAIT)
dmu_tx_wait(tx);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, &xoid, tx, cr, IS_XATTR, &xzp, 0);
ASSERT(xzp->z_id == xoid);
ASSERT(xzp->z_phys->zp_parent == zp->z_id);
dmu_buf_will_dirty(zp->z_dbuf, tx);
zp->z_phys->zp_xattr = xoid;
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp, xzp, "");
dmu_tx_commit(tx);
#ifdef __APPLE__
/*
* Obtain and attach the vnode after committing the transaction
*/
zfs_attach_vnode(xzp);
#endif
*xvpp = ZTOV(xzp);
return (0);
}
static void
dsl_deleg_set_sync(void *arg, dmu_tx_t *tx)
{
dsl_deleg_arg_t *dda = arg;
dsl_dir_t *dd;
dsl_pool_t *dp = dmu_tx_pool(tx);
objset_t *mos = dp->dp_meta_objset;
nvpair_t *whopair = NULL;
uint64_t zapobj;
VERIFY0(dsl_dir_hold(dp, dda->dda_name, FTAG, &dd, NULL));
zapobj = dd->dd_phys->dd_deleg_zapobj;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
while ((whopair = nvlist_next_nvpair(dda->dda_nvlist, whopair))) {
const char *whokey = nvpair_name(whopair);
nvlist_t *perms;
nvpair_t *permpair = NULL;
uint64_t jumpobj;
perms = fnvpair_value_nvlist(whopair);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) != 0) {
jumpobj = zap_create_link(mos, DMU_OT_DSL_PERMS,
zapobj, whokey, tx);
}
while ((permpair = nvlist_next_nvpair(perms, permpair))) {
const char *perm = nvpair_name(permpair);
uint64_t n = 0;
VERIFY(zap_update(mos, jumpobj,
perm, 8, 1, &n, tx) == 0);
spa_history_log_internal_dd(dd, "permission update", tx,
"%s %s", whokey, perm);
}
}
dsl_dir_rele(dd, FTAG);
}
static void
dsl_deleg_set_sync(void *arg1, void *arg2, cred_t *cr, dmu_tx_t *tx)
{
dsl_dir_t *dd = arg1;
nvlist_t *nvp = arg2;
objset_t *mos = dd->dd_pool->dp_meta_objset;
nvpair_t *whopair = NULL;
uint64_t zapobj = dd->dd_phys->dd_deleg_zapobj;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
while (whopair = nvlist_next_nvpair(nvp, whopair)) {
const char *whokey = nvpair_name(whopair);
nvlist_t *perms;
nvpair_t *permpair = NULL;
uint64_t jumpobj;
VERIFY(nvpair_value_nvlist(whopair, &perms) == 0);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) != 0) {
jumpobj = zap_create(mos, DMU_OT_DSL_PERMS,
DMU_OT_NONE, 0, tx);
VERIFY(zap_update(mos, zapobj,
whokey, 8, 1, &jumpobj, tx) == 0);
}
while (permpair = nvlist_next_nvpair(perms, permpair)) {
const char *perm = nvpair_name(permpair);
uint64_t n = 0;
VERIFY(zap_update(mos, jumpobj,
perm, 8, 1, &n, tx) == 0);
spa_history_internal_log(LOG_DS_PERM_UPDATE,
dd->dd_pool->dp_spa, tx, cr,
"%s %s dataset = %llu", whokey, perm,
dd->dd_phys->dd_head_dataset_obj);
}
}
}
static void
copy_create_perms(dsl_dir_t *dd, uint64_t pzapobj,
boolean_t dosets, uint64_t uid, dmu_tx_t *tx)
{
objset_t *mos = dd->dd_pool->dp_meta_objset;
uint64_t jumpobj, pjumpobj;
uint64_t zapobj = dd->dd_phys->dd_deleg_zapobj;
zap_cursor_t zc;
zap_attribute_t za;
char whokey[ZFS_MAX_DELEG_NAME];
zfs_deleg_whokey(whokey,
dosets ? ZFS_DELEG_CREATE_SETS : ZFS_DELEG_CREATE,
ZFS_DELEG_LOCAL, NULL);
if (zap_lookup(mos, pzapobj, whokey, 8, 1, &pjumpobj) != 0)
return;
if (zapobj == 0) {
dmu_buf_will_dirty(dd->dd_dbuf, tx);
zapobj = dd->dd_phys->dd_deleg_zapobj = zap_create(mos,
DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
}
zfs_deleg_whokey(whokey,
dosets ? ZFS_DELEG_USER_SETS : ZFS_DELEG_USER,
ZFS_DELEG_LOCAL, &uid);
if (zap_lookup(mos, zapobj, whokey, 8, 1, &jumpobj) == ENOENT) {
jumpobj = zap_create(mos, DMU_OT_DSL_PERMS, DMU_OT_NONE, 0, tx);
VERIFY(zap_add(mos, zapobj, whokey, 8, 1, &jumpobj, tx) == 0);
}
for (zap_cursor_init(&zc, mos, pjumpobj);
zap_cursor_retrieve(&zc, &za) == 0;
zap_cursor_advance(&zc)) {
uint64_t zero = 0;
ASSERT(za.za_integer_length == 8 && za.za_num_integers == 1);
VERIFY(zap_update(mos, jumpobj, za.za_name,
8, 1, &zero, tx) == 0);
}
zap_cursor_fini(&zc);
}
static void
dsl_bookmark_destroy_sync(void *arg, dmu_tx_t *tx)
{
dsl_bookmark_destroy_arg_t *dbda = arg;
dsl_pool_t *dp = dmu_tx_pool(tx);
objset_t *mos = dp->dp_meta_objset;
nvpair_t *pair;
for (pair = nvlist_next_nvpair(dbda->dbda_success, NULL);
pair != NULL; pair = nvlist_next_nvpair(dbda->dbda_success, pair)) {
dsl_dataset_t *ds;
char *shortname;
uint64_t zap_cnt;
VERIFY0(dsl_bookmark_hold_ds(dp, nvpair_name(pair),
&ds, FTAG, &shortname));
VERIFY0(dsl_dataset_bookmark_remove(ds, shortname, tx));
/*
* If all of this dataset's bookmarks have been destroyed,
* free the zap object and decrement the feature's use count.
*/
VERIFY0(zap_count(mos, ds->ds_bookmarks,
&zap_cnt));
if (zap_cnt == 0) {
dmu_buf_will_dirty(ds->ds_dbuf, tx);
VERIFY0(zap_destroy(mos, ds->ds_bookmarks, tx));
ds->ds_bookmarks = 0;
spa_feature_decr(dp->dp_spa, SPA_FEATURE_BOOKMARKS, tx);
VERIFY0(zap_remove(mos, ds->ds_object,
DS_FIELD_BOOKMARK_NAMES, tx));
}
spa_history_log_internal_ds(ds, "remove bookmark", tx,
"name=%s", shortname);
dsl_dataset_rele(ds, FTAG);
}
}
int
zfs_make_xattrdir(znode_t *zp, vattr_t *vap, vnode_t **xvpp, cred_t *cr)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
znode_t *xzp;
dmu_tx_t *tx;
uint64_t xoid;
int error;
*xvpp = NULL;
if (error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, cr))
return (error);
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_bonus(tx, zp->z_id);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
error = dmu_tx_assign(tx, zfsvfs->z_assign);
if (error) {
if (error == ERESTART && zfsvfs->z_assign == TXG_NOWAIT)
dmu_tx_wait(tx);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, &xoid, tx, cr, IS_XATTR, &xzp, 0);
ASSERT(xzp->z_id == xoid);
ASSERT(xzp->z_phys->zp_parent == zp->z_id);
dmu_buf_will_dirty(zp->z_dbuf, tx);
zp->z_phys->zp_xattr = xoid;
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp, xzp, "");
dmu_tx_commit(tx);
*xvpp = ZTOV(xzp);
return (0);
}
uint64_t
dsl_dir_create_sync(dsl_pool_t *dp, dsl_dir_t *pds, const char *name,
dmu_tx_t *tx)
{
objset_t *mos = dp->dp_meta_objset;
uint64_t ddobj;
dsl_dir_phys_t *ddphys;
dmu_buf_t *dbuf;
ddobj = dmu_object_alloc(mos, DMU_OT_DSL_DIR, 0,
DMU_OT_DSL_DIR, sizeof (dsl_dir_phys_t), tx);
if (pds) {
VERIFY(0 == zap_add(mos, pds->dd_phys->dd_child_dir_zapobj,
name, sizeof (uint64_t), 1, &ddobj, tx));
} else {
/* it's the root dir */
VERIFY(0 == zap_add(mos, DMU_POOL_DIRECTORY_OBJECT,
DMU_POOL_ROOT_DATASET, sizeof (uint64_t), 1, &ddobj, tx));
}
VERIFY(0 == dmu_bonus_hold(mos, ddobj, FTAG, &dbuf));
dmu_buf_will_dirty(dbuf, tx);
ddphys = dbuf->db_data;
ddphys->dd_creation_time = gethrestime_sec();
if (pds)
ddphys->dd_parent_obj = pds->dd_object;
ddphys->dd_props_zapobj = zap_create(mos,
DMU_OT_DSL_PROPS, DMU_OT_NONE, 0, tx);
ddphys->dd_child_dir_zapobj = zap_create(mos,
DMU_OT_DSL_DIR_CHILD_MAP, DMU_OT_NONE, 0, tx);
if (spa_version(dp->dp_spa) >= SPA_VERSION_USED_BREAKDOWN)
ddphys->dd_flags |= DD_FLAG_USED_BREAKDOWN;
dmu_buf_rele(dbuf, FTAG);
return (ddobj);
}
static int
zap_table_grow(zap_t *zap, zap_table_phys_t *tbl,
void (*transfer_func)(const uint64_t *src, uint64_t *dst, int n),
dmu_tx_t *tx)
{
uint64_t b, newblk;
dmu_buf_t *db_old, *db_new;
int err;
int bs = FZAP_BLOCK_SHIFT(zap);
int hepb = 1<<(bs-4);
/* hepb = half the number of entries in a block */
ASSERT(RW_WRITE_HELD(&zap->zap_rwlock));
ASSERT(tbl->zt_blk != 0);
ASSERT(tbl->zt_numblks > 0);
if (tbl->zt_nextblk != 0) {
newblk = tbl->zt_nextblk;
} else {
newblk = zap_allocate_blocks(zap, tbl->zt_numblks * 2);
tbl->zt_nextblk = newblk;
ASSERT0(tbl->zt_blks_copied);
dmu_prefetch(zap->zap_objset, zap->zap_object,
tbl->zt_blk << bs, tbl->zt_numblks << bs);
}
/*
* Copy the ptrtbl from the old to new location.
*/
b = tbl->zt_blks_copied;
err = dmu_buf_hold(zap->zap_objset, zap->zap_object,
(tbl->zt_blk + b) << bs, FTAG, &db_old, DMU_READ_NO_PREFETCH);
if (err)
return (err);
/* first half of entries in old[b] go to new[2*b+0] */
VERIFY(0 == dmu_buf_hold(zap->zap_objset, zap->zap_object,
(newblk + 2*b+0) << bs, FTAG, &db_new, DMU_READ_NO_PREFETCH));
dmu_buf_will_dirty(db_new, tx);
transfer_func(db_old->db_data, db_new->db_data, hepb);
dmu_buf_rele(db_new, FTAG);
/* second half of entries in old[b] go to new[2*b+1] */
VERIFY(0 == dmu_buf_hold(zap->zap_objset, zap->zap_object,
(newblk + 2*b+1) << bs, FTAG, &db_new, DMU_READ_NO_PREFETCH));
dmu_buf_will_dirty(db_new, tx);
transfer_func((uint64_t *)db_old->db_data + hepb,
db_new->db_data, hepb);
dmu_buf_rele(db_new, FTAG);
dmu_buf_rele(db_old, FTAG);
tbl->zt_blks_copied++;
dprintf("copied block %llu of %llu\n",
tbl->zt_blks_copied, tbl->zt_numblks);
if (tbl->zt_blks_copied == tbl->zt_numblks) {
(void) dmu_free_range(zap->zap_objset, zap->zap_object,
tbl->zt_blk << bs, tbl->zt_numblks << bs, tx);
tbl->zt_blk = newblk;
tbl->zt_numblks *= 2;
tbl->zt_shift++;
tbl->zt_nextblk = 0;
tbl->zt_blks_copied = 0;
dprintf("finished; numblocks now %llu (%lluk entries)\n",
tbl->zt_numblks, 1<<(tbl->zt_shift-10));
}
return (0);
}
/*ARGSUSED*/
static void
spa_history_log_sync(void *arg, dmu_tx_t *tx)
{
nvlist_t *nvl = arg;
spa_t *spa = dmu_tx_pool(tx)->dp_spa;
objset_t *mos = spa->spa_meta_objset;
dmu_buf_t *dbp;
spa_history_phys_t *shpp;
size_t reclen;
uint64_t le_len;
char *record_packed = NULL;
int ret;
/*
* If we have an older pool that doesn't have a command
* history object, create it now.
*/
mutex_enter(&spa->spa_history_lock);
if (!spa->spa_history)
spa_history_create_obj(spa, tx);
mutex_exit(&spa->spa_history_lock);
/*
* Get the offset of where we need to write via the bonus buffer.
* Update the offset when the write completes.
*/
VERIFY0(dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp));
shpp = dbp->db_data;
dmu_buf_will_dirty(dbp, tx);
#ifdef ZFS_DEBUG
{
dmu_object_info_t doi;
dmu_object_info_from_db(dbp, &doi);
ASSERT3U(doi.doi_bonus_type, ==, DMU_OT_SPA_HISTORY_OFFSETS);
}
#endif
fnvlist_add_uint64(nvl, ZPOOL_HIST_TIME, gethrestime_sec());
fnvlist_add_string(nvl, ZPOOL_HIST_HOST, utsname()->nodename);
if (nvlist_exists(nvl, ZPOOL_HIST_CMD)) {
zfs_dbgmsg("command: %s",
fnvlist_lookup_string(nvl, ZPOOL_HIST_CMD));
} else if (nvlist_exists(nvl, ZPOOL_HIST_INT_NAME)) {
if (nvlist_exists(nvl, ZPOOL_HIST_DSNAME)) {
zfs_dbgmsg("txg %lld %s %s (id %llu) %s",
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME),
fnvlist_lookup_string(nvl, ZPOOL_HIST_DSNAME),
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_DSID),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR));
} else {
zfs_dbgmsg("txg %lld %s %s",
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR));
}
/*
* The history sysevent is posted only for internal history
* messages to show what has happened, not how it happened. For
* example, the following command:
*
* # zfs destroy -r tank/foo
*
* will result in one sysevent posted per dataset that is
* destroyed as a result of the command - which could be more
* than one event in total. By contrast, if the sysevent was
* posted as a result of the ZPOOL_HIST_CMD key being present
* it would result in only one sysevent being posted with the
* full command line arguments, requiring the consumer to know
* how to parse and understand zfs(1M) command invocations.
*/
spa_history_log_notify(spa, nvl);
} else if (nvlist_exists(nvl, ZPOOL_HIST_IOCTL)) {
zfs_dbgmsg("ioctl %s",
fnvlist_lookup_string(nvl, ZPOOL_HIST_IOCTL));
}
VERIFY3U(nvlist_pack(nvl, &record_packed, &reclen, NV_ENCODE_NATIVE,
KM_SLEEP), ==, 0);
mutex_enter(&spa->spa_history_lock);
/* write out the packed length as little endian */
le_len = LE_64((uint64_t)reclen);
ret = spa_history_write(spa, &le_len, sizeof (le_len), shpp, tx);
if (!ret)
ret = spa_history_write(spa, record_packed, reclen, shpp, tx);
/* The first command is the create, which we keep forever */
if (ret == 0 && shpp->sh_pool_create_len == 0 &&
nvlist_exists(nvl, ZPOOL_HIST_CMD)) {
shpp->sh_pool_create_len = shpp->sh_bof = shpp->sh_eof;
}
mutex_exit(&spa->spa_history_lock);
fnvlist_pack_free(record_packed, reclen);
dmu_buf_rele(dbp, FTAG);
fnvlist_free(nvl);
}
void
zfs_rmnode(znode_t *zp)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
objset_t *os = zfsvfs->z_os;
znode_t *xzp = NULL;
dmu_tx_t *tx;
uint64_t acl_obj;
int error;
int vfslocked;
vfslocked = VFS_LOCK_GIANT(zfsvfs->z_vfs);
ASSERT(zp->z_phys->zp_links == 0);
/*
* If this is a ZIL replay then leave the object in the unlinked set.
* Otherwise we can get a deadlock, because the delete can be
* quite large and span multiple tx's and txgs, but each replay
* creates a tx to atomically run the replay function and mark the
* replay record as complete. We deadlock trying to start a tx in
* a new txg to further the deletion but can't because the replay
* tx hasn't finished.
*
* We actually delete the object if we get a failure to create an
* object in zil_replay_log_record(), or after calling zil_replay().
*/
if (zfsvfs->z_assign >= TXG_INITIAL) {
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
return;
}
/*
* If this is an attribute directory, purge its contents.
*/
if (ZTOV(zp) != NULL && ZTOV(zp)->v_type == VDIR &&
(zp->z_phys->zp_flags & ZFS_XATTR)) {
if (zfs_purgedir(zp) != 0) {
/*
* Not enough space to delete some xattrs.
* Leave it in the unlinked set.
*/
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
VFS_UNLOCK_GIANT(vfslocked);
return;
}
}
/*
* Free up all the data in the file.
*/
error = dmu_free_long_range(os, zp->z_id, 0, DMU_OBJECT_END);
if (error) {
/*
* Not enough space. Leave the file in the unlinked set.
*/
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
return;
}
/*
* If the file has extended attributes, we're going to unlink
* the xattr dir.
*/
if (zp->z_phys->zp_xattr) {
error = zfs_zget(zfsvfs, zp->z_phys->zp_xattr, &xzp);
ASSERT(error == 0);
}
acl_obj = zp->z_phys->zp_acl.z_acl_extern_obj;
/*
* Set up the final transaction.
*/
tx = dmu_tx_create(os);
dmu_tx_hold_free(tx, zp->z_id, 0, DMU_OBJECT_END);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
if (xzp) {
dmu_tx_hold_bonus(tx, xzp->z_id);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, TRUE, NULL);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
/*
* Not enough space to delete the file. Leave it in the
* unlinked set, leaking it until the fs is remounted (at
* which point we'll call zfs_unlinked_drain() to process it).
*/
dmu_tx_abort(tx);
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
goto out;
}
if (xzp) {
dmu_buf_will_dirty(xzp->z_dbuf, tx);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE; /* mark xzp for deletion */
xzp->z_phys->zp_links = 0; /* no more links to it */
mutex_exit(&xzp->z_lock);
zfs_unlinked_add(xzp, tx);
}
/* Remove this znode from the unlinked set */
VERIFY3U(0, ==,
zap_remove_int(zfsvfs->z_os, zfsvfs->z_unlinkedobj, zp->z_id, tx));
zfs_znode_delete(zp, tx);
dmu_tx_commit(tx);
out:
if (xzp)
VN_RELE(ZTOV(xzp));
VFS_UNLOCK_GIANT(vfslocked);
}
/*ARGSUSED*/
static void
spa_history_log_sync(void *arg1, void *arg2, dmu_tx_t *tx)
{
spa_t *spa = arg1;
history_arg_t *hap = arg2;
const char *history_str = hap->ha_history_str;
objset_t *mos = spa->spa_meta_objset;
dmu_buf_t *dbp;
spa_history_phys_t *shpp;
size_t reclen;
uint64_t le_len;
nvlist_t *nvrecord;
char *record_packed = NULL;
int ret;
/*
* If we have an older pool that doesn't have a command
* history object, create it now.
*/
mutex_enter(&spa->spa_history_lock);
if (!spa->spa_history)
spa_history_create_obj(spa, tx);
mutex_exit(&spa->spa_history_lock);
/*
* Get the offset of where we need to write via the bonus buffer.
* Update the offset when the write completes.
*/
VERIFY(0 == dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp));
shpp = dbp->db_data;
dmu_buf_will_dirty(dbp, tx);
#ifdef ZFS_DEBUG
{
dmu_object_info_t doi;
dmu_object_info_from_db(dbp, &doi);
ASSERT3U(doi.doi_bonus_type, ==, DMU_OT_SPA_HISTORY_OFFSETS);
}
#endif
VERIFY(nvlist_alloc(&nvrecord, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_TIME,
gethrestime_sec()) == 0);
VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_WHO, hap->ha_uid) == 0);
if (hap->ha_zone != NULL)
VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_ZONE,
hap->ha_zone) == 0);
#ifdef _KERNEL
VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_HOST,
utsname.nodename) == 0);
#endif
if (hap->ha_log_type == LOG_CMD_POOL_CREATE ||
hap->ha_log_type == LOG_CMD_NORMAL) {
VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_CMD,
history_str) == 0);
zfs_dbgmsg("command: %s", history_str);
} else {
VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_INT_EVENT,
hap->ha_event) == 0);
VERIFY(nvlist_add_uint64(nvrecord, ZPOOL_HIST_TXG,
tx->tx_txg) == 0);
VERIFY(nvlist_add_string(nvrecord, ZPOOL_HIST_INT_STR,
history_str) == 0);
zfs_dbgmsg("internal %s pool:%s txg:%llu %s",
zfs_history_event_names[hap->ha_event], spa_name(spa),
(longlong_t)tx->tx_txg, history_str);
}
VERIFY(nvlist_size(nvrecord, &reclen, NV_ENCODE_XDR) == 0);
record_packed = kmem_alloc(reclen, KM_SLEEP);
VERIFY(nvlist_pack(nvrecord, &record_packed, &reclen,
NV_ENCODE_XDR, KM_SLEEP) == 0);
mutex_enter(&spa->spa_history_lock);
if (hap->ha_log_type == LOG_CMD_POOL_CREATE)
VERIFY(shpp->sh_eof == shpp->sh_pool_create_len);
/* write out the packed length as little endian */
le_len = LE_64((uint64_t)reclen);
ret = spa_history_write(spa, &le_len, sizeof (le_len), shpp, tx);
if (!ret)
ret = spa_history_write(spa, record_packed, reclen, shpp, tx);
if (!ret && hap->ha_log_type == LOG_CMD_POOL_CREATE) {
shpp->sh_pool_create_len += sizeof (le_len) + reclen;
shpp->sh_bof = shpp->sh_pool_create_len;
}
mutex_exit(&spa->spa_history_lock);
nvlist_free(nvrecord);
kmem_free(record_packed, reclen);
dmu_buf_rele(dbp, FTAG);
strfree(hap->ha_history_str);
if (hap->ha_zone != NULL)
strfree(hap->ha_zone);
kmem_free(hap, sizeof (history_arg_t));
}
void
dsl_destroy_snapshot_sync_impl(dsl_dataset_t *ds, boolean_t defer, dmu_tx_t *tx)
{
#ifdef ZFS_DEBUG
int err;
#endif
spa_feature_t f;
int after_branch_point = FALSE;
dsl_pool_t *dp = ds->ds_dir->dd_pool;
objset_t *mos = dp->dp_meta_objset;
dsl_dataset_t *ds_prev = NULL;
uint64_t obj, old_unique, used = 0, comp = 0, uncomp = 0;
dsl_dataset_t *ds_next, *ds_head, *hds;
ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock));
ASSERT3U(dsl_dataset_phys(ds)->ds_bp.blk_birth, <=, tx->tx_txg);
ASSERT(refcount_is_zero(&ds->ds_longholds));
if (defer &&
(ds->ds_userrefs > 0 ||
dsl_dataset_phys(ds)->ds_num_children > 1)) {
ASSERT(spa_version(dp->dp_spa) >= SPA_VERSION_USERREFS);
dmu_buf_will_dirty(ds->ds_dbuf, tx);
dsl_dataset_phys(ds)->ds_flags |= DS_FLAG_DEFER_DESTROY;
spa_history_log_internal_ds(ds, "defer_destroy", tx, "");
return;
}
ASSERT3U(dsl_dataset_phys(ds)->ds_num_children, <=, 1);
/* We need to log before removing it from the namespace. */
spa_history_log_internal_ds(ds, "destroy", tx, "");
dsl_scan_ds_destroyed(ds, tx);
obj = ds->ds_object;
for (f = 0; f < SPA_FEATURES; f++) {
if (ds->ds_feature_inuse[f]) {
dsl_dataset_deactivate_feature(obj, f, tx);
ds->ds_feature_inuse[f] = B_FALSE;
}
}
if (dsl_dataset_phys(ds)->ds_prev_snap_obj != 0) {
ASSERT3P(ds->ds_prev, ==, NULL);
VERIFY0(dsl_dataset_hold_obj(dp,
dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &ds_prev));
after_branch_point =
(dsl_dataset_phys(ds_prev)->ds_next_snap_obj != obj);
dmu_buf_will_dirty(ds_prev->ds_dbuf, tx);
if (after_branch_point &&
dsl_dataset_phys(ds_prev)->ds_next_clones_obj != 0) {
dsl_dataset_remove_from_next_clones(ds_prev, obj, tx);
if (dsl_dataset_phys(ds)->ds_next_snap_obj != 0) {
VERIFY0(zap_add_int(mos,
dsl_dataset_phys(ds_prev)->
ds_next_clones_obj,
dsl_dataset_phys(ds)->ds_next_snap_obj,
tx));
}
}
if (!after_branch_point) {
dsl_dataset_phys(ds_prev)->ds_next_snap_obj =
dsl_dataset_phys(ds)->ds_next_snap_obj;
}
}
void
zfs_rmnode(znode_t *zp)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
objset_t *os = zfsvfs->z_os;
znode_t *xzp = NULL;
dmu_tx_t *tx;
uint64_t acl_obj;
int error;
ASSERT(ZTOV(zp)->v_count == 0);
ASSERT(zp->z_phys->zp_links == 0);
/*
* If this is an attribute directory, purge its contents.
*/
if (ZTOV(zp)->v_type == VDIR && (zp->z_phys->zp_flags & ZFS_XATTR)) {
if (zfs_purgedir(zp) != 0) {
/*
* Not enough space to delete some xattrs.
* Leave it in the unlinked set.
*/
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
return;
}
}
/*
* Free up all the data in the file.
*/
error = dmu_free_long_range(os, zp->z_id, 0, DMU_OBJECT_END);
if (error) {
/*
* Not enough space. Leave the file in the unlinked set.
*/
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
return;
}
/*
* If the file has extended attributes, we're going to unlink
* the xattr dir.
*/
if (zp->z_phys->zp_xattr) {
error = zfs_zget(zfsvfs, zp->z_phys->zp_xattr, &xzp);
ASSERT(error == 0);
}
acl_obj = zp->z_phys->zp_acl.z_acl_extern_obj;
/*
* Set up the final transaction.
*/
tx = dmu_tx_create(os);
dmu_tx_hold_free(tx, zp->z_id, 0, DMU_OBJECT_END);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
if (xzp) {
dmu_tx_hold_bonus(tx, xzp->z_id);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, TRUE, NULL);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
/*
* Not enough space to delete the file. Leave it in the
* unlinked set, leaking it until the fs is remounted (at
* which point we'll call zfs_unlinked_drain() to process it).
*/
dmu_tx_abort(tx);
zfs_znode_dmu_fini(zp);
zfs_znode_free(zp);
goto out;
}
if (xzp) {
dmu_buf_will_dirty(xzp->z_dbuf, tx);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE; /* mark xzp for deletion */
xzp->z_phys->zp_links = 0; /* no more links to it */
mutex_exit(&xzp->z_lock);
zfs_unlinked_add(xzp, tx);
}
/* Remove this znode from the unlinked set */
VERIFY3U(0, ==,
zap_remove_int(zfsvfs->z_os, zfsvfs->z_unlinkedobj, zp->z_id, tx));
zfs_znode_delete(zp, tx);
dmu_tx_commit(tx);
out:
if (xzp)
VN_RELE(ZTOV(xzp));
}
/*ARGSUSED*/
static void
spa_history_log_sync(void *arg, dmu_tx_t *tx)
{
nvlist_t *nvl = arg;
spa_t *spa = dmu_tx_pool(tx)->dp_spa;
objset_t *mos = spa->spa_meta_objset;
dmu_buf_t *dbp;
spa_history_phys_t *shpp;
size_t reclen;
uint64_t le_len;
char *record_packed = NULL;
int ret;
/*
* If we have an older pool that doesn't have a command
* history object, create it now.
*/
mutex_enter(&spa->spa_history_lock);
if (!spa->spa_history)
spa_history_create_obj(spa, tx);
mutex_exit(&spa->spa_history_lock);
/*
* Get the offset of where we need to write via the bonus buffer.
* Update the offset when the write completes.
*/
VERIFY0(dmu_bonus_hold(mos, spa->spa_history, FTAG, &dbp));
shpp = dbp->db_data;
dmu_buf_will_dirty(dbp, tx);
#ifdef ZFS_DEBUG
{
dmu_object_info_t doi;
dmu_object_info_from_db(dbp, &doi);
ASSERT3U(doi.doi_bonus_type, ==, DMU_OT_SPA_HISTORY_OFFSETS);
}
#endif
fnvlist_add_uint64(nvl, ZPOOL_HIST_TIME, gethrestime_sec());
#ifdef _KERNEL
fnvlist_add_string(nvl, ZPOOL_HIST_HOST, utsname.nodename);
#endif
if (nvlist_exists(nvl, ZPOOL_HIST_CMD)) {
zfs_dbgmsg("command: %s",
fnvlist_lookup_string(nvl, ZPOOL_HIST_CMD));
} else if (nvlist_exists(nvl, ZPOOL_HIST_INT_NAME)) {
if (nvlist_exists(nvl, ZPOOL_HIST_DSNAME)) {
zfs_dbgmsg("txg %lld %s %s (id %llu) %s",
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME),
fnvlist_lookup_string(nvl, ZPOOL_HIST_DSNAME),
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_DSID),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR));
} else {
zfs_dbgmsg("txg %lld %s %s",
fnvlist_lookup_uint64(nvl, ZPOOL_HIST_TXG),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_NAME),
fnvlist_lookup_string(nvl, ZPOOL_HIST_INT_STR));
}
} else if (nvlist_exists(nvl, ZPOOL_HIST_IOCTL)) {
zfs_dbgmsg("ioctl %s",
fnvlist_lookup_string(nvl, ZPOOL_HIST_IOCTL));
}
record_packed = fnvlist_pack(nvl, &reclen);
mutex_enter(&spa->spa_history_lock);
/* write out the packed length as little endian */
le_len = LE_64((uint64_t)reclen);
ret = spa_history_write(spa, &le_len, sizeof (le_len), shpp, tx);
if (!ret)
ret = spa_history_write(spa, record_packed, reclen, shpp, tx);
/* The first command is the create, which we keep forever */
if (ret == 0 && shpp->sh_pool_create_len == 0 &&
nvlist_exists(nvl, ZPOOL_HIST_CMD)) {
shpp->sh_pool_create_len = shpp->sh_bof = shpp->sh_eof;
}
mutex_exit(&spa->spa_history_lock);
fnvlist_pack_free(record_packed, reclen);
dmu_buf_rele(dbp, FTAG);
fnvlist_free(nvl);
}
/*
* sync out AVL trees to persistent storage.
*/
void
zfs_fuid_sync(zfsvfs_t *zfsvfs, dmu_tx_t *tx)
{
#ifdef HAVE_ZPL
nvlist_t *nvp;
nvlist_t **fuids;
size_t nvsize = 0;
char *packed;
dmu_buf_t *db;
fuid_domain_t *domnode;
int numnodes;
int i;
if (!zfsvfs->z_fuid_dirty) {
return;
}
rw_enter(&zfsvfs->z_fuid_lock, RW_WRITER);
/*
* First see if table needs to be created?
*/
if (zfsvfs->z_fuid_obj == 0) {
zfsvfs->z_fuid_obj = dmu_object_alloc(zfsvfs->z_os,
DMU_OT_FUID, 1 << 14, DMU_OT_FUID_SIZE,
sizeof (uint64_t), tx);
VERIFY(zap_add(zfsvfs->z_os, MASTER_NODE_OBJ,
ZFS_FUID_TABLES, sizeof (uint64_t), 1,
&zfsvfs->z_fuid_obj, tx) == 0);
}
VERIFY(nvlist_alloc(&nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0);
numnodes = avl_numnodes(&zfsvfs->z_fuid_idx);
fuids = kmem_alloc(numnodes * sizeof (void *), KM_SLEEP);
for (i = 0, domnode = avl_first(&zfsvfs->z_fuid_domain); domnode; i++,
domnode = AVL_NEXT(&zfsvfs->z_fuid_domain, domnode)) {
VERIFY(nvlist_alloc(&fuids[i], NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(fuids[i], FUID_IDX,
domnode->f_idx) == 0);
VERIFY(nvlist_add_uint64(fuids[i], FUID_OFFSET, 0) == 0);
VERIFY(nvlist_add_string(fuids[i], FUID_DOMAIN,
domnode->f_ksid->kd_name) == 0);
}
VERIFY(nvlist_add_nvlist_array(nvp, FUID_NVP_ARRAY,
fuids, numnodes) == 0);
for (i = 0; i != numnodes; i++)
nvlist_free(fuids[i]);
kmem_free(fuids, numnodes * sizeof (void *));
VERIFY(nvlist_size(nvp, &nvsize, NV_ENCODE_XDR) == 0);
packed = kmem_alloc(nvsize, KM_SLEEP);
VERIFY(nvlist_pack(nvp, &packed, &nvsize,
NV_ENCODE_XDR, KM_SLEEP) == 0);
nvlist_free(nvp);
zfsvfs->z_fuid_size = nvsize;
dmu_write(zfsvfs->z_os, zfsvfs->z_fuid_obj, 0,
zfsvfs->z_fuid_size, packed, tx);
kmem_free(packed, zfsvfs->z_fuid_size);
VERIFY(0 == dmu_bonus_hold(zfsvfs->z_os, zfsvfs->z_fuid_obj,
FTAG, &db));
dmu_buf_will_dirty(db, tx);
*(uint64_t *)db->db_data = zfsvfs->z_fuid_size;
dmu_buf_rele(db, FTAG);
zfsvfs->z_fuid_dirty = B_FALSE;
rw_exit(&zfsvfs->z_fuid_lock);
#endif /* HAVE_ZPL */
}
/*
* Main attribute lookup/update function
* returns 0 for success or non zero for failures
*
* Operates on bulk array, first failure will abort further processing
*/
int
sa_attr_op(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count,
sa_data_op_t data_op, dmu_tx_t *tx)
{
sa_os_t *sa = hdl->sa_os->os_sa;
int i;
int error = 0;
sa_buf_type_t buftypes;
buftypes = 0;
ASSERT(count > 0);
for (i = 0; i != count; i++) {
ASSERT(bulk[i].sa_attr <= hdl->sa_os->os_sa->sa_num_attrs);
bulk[i].sa_addr = NULL;
/* First check the bonus buffer */
if (hdl->sa_bonus_tab && TOC_ATTR_PRESENT(
hdl->sa_bonus_tab->sa_idx_tab[bulk[i].sa_attr])) {
SA_ATTR_INFO(sa, hdl->sa_bonus_tab,
SA_GET_HDR(hdl, SA_BONUS),
bulk[i].sa_attr, bulk[i], SA_BONUS, hdl);
if (tx && !(buftypes & SA_BONUS)) {
dmu_buf_will_dirty(hdl->sa_bonus, tx);
buftypes |= SA_BONUS;
}
}
if (bulk[i].sa_addr == NULL &&
((error = sa_get_spill(hdl)) == 0)) {
if (TOC_ATTR_PRESENT(
hdl->sa_spill_tab->sa_idx_tab[bulk[i].sa_attr])) {
SA_ATTR_INFO(sa, hdl->sa_spill_tab,
SA_GET_HDR(hdl, SA_SPILL),
bulk[i].sa_attr, bulk[i], SA_SPILL, hdl);
if (tx && !(buftypes & SA_SPILL) &&
bulk[i].sa_size == bulk[i].sa_length) {
dmu_buf_will_dirty(hdl->sa_spill, tx);
buftypes |= SA_SPILL;
}
}
}
if (error && error != ENOENT) {
return ((error == ECKSUM) ? EIO : error);
}
switch (data_op) {
case SA_LOOKUP:
if (bulk[i].sa_addr == NULL)
return (ENOENT);
if (bulk[i].sa_data) {
SA_COPY_DATA(bulk[i].sa_data_func,
bulk[i].sa_addr, bulk[i].sa_data,
bulk[i].sa_size);
}
continue;
case SA_UPDATE:
/* existing rewrite of attr */
if (bulk[i].sa_addr &&
bulk[i].sa_size == bulk[i].sa_length) {
SA_COPY_DATA(bulk[i].sa_data_func,
bulk[i].sa_data, bulk[i].sa_addr,
bulk[i].sa_length);
continue;
} else if (bulk[i].sa_addr) { /* attr size change */
error = sa_modify_attrs(hdl, bulk[i].sa_attr,
SA_REPLACE, bulk[i].sa_data_func,
bulk[i].sa_data, bulk[i].sa_length, tx);
} else { /* adding new attribute */
error = sa_modify_attrs(hdl, bulk[i].sa_attr,
SA_ADD, bulk[i].sa_data_func,
bulk[i].sa_data, bulk[i].sa_length, tx);
}
if (error)
return (error);
break;
}
}
return (error);
}
/*
* Unlink zp from dl, and mark zp for deletion if this was the last link.
* Can fail if zp is a mount point (EBUSY) or a non-empty directory (EEXIST).
* If 'unlinkedp' is NULL, we put unlinked znodes on the unlinked list.
* If it's non-NULL, we use it to indicate whether the znode needs deletion,
* and it's the caller's job to do it.
*/
int
zfs_link_destroy(zfs_dirlock_t *dl, znode_t *zp, dmu_tx_t *tx, int flag,
boolean_t *unlinkedp)
{
znode_t *dzp = dl->dl_dzp;
vnode_t *vp = ZTOV(zp);
int zp_is_dir = (vp->v_type == VDIR);
boolean_t unlinked = B_FALSE;
int error;
dnlc_remove(ZTOV(dzp), dl->dl_name);
if (!(flag & ZRENAMING)) {
dmu_buf_will_dirty(zp->z_dbuf, tx);
if (vn_vfswlock(vp)) /* prevent new mounts on zp */
return (EBUSY);
if (vn_ismntpt(vp)) { /* don't remove mount point */
vn_vfsunlock(vp);
return (EBUSY);
}
mutex_enter(&zp->z_lock);
if (zp_is_dir && !zfs_dirempty(zp)) { /* dir not empty */
mutex_exit(&zp->z_lock);
vn_vfsunlock(vp);
return (ENOTEMPTY);
}
if (zp->z_phys->zp_links <= zp_is_dir) {
zfs_panic_recover("zfs: link count on vnode %p is %u, "
"should be at least %u", zp->z_vnode,
(int)zp->z_phys->zp_links,
zp_is_dir + 1);
zp->z_phys->zp_links = zp_is_dir + 1;
}
if (--zp->z_phys->zp_links == zp_is_dir) {
zp->z_unlinked = B_TRUE;
zp->z_phys->zp_links = 0;
unlinked = B_TRUE;
} else {
zfs_time_stamper_locked(zp, STATE_CHANGED, tx);
}
mutex_exit(&zp->z_lock);
vn_vfsunlock(vp);
}
dmu_buf_will_dirty(dzp->z_dbuf, tx);
mutex_enter(&dzp->z_lock);
dzp->z_phys->zp_size--; /* one dirent removed */
dzp->z_phys->zp_links -= zp_is_dir; /* ".." link from zp */
zfs_time_stamper_locked(dzp, CONTENT_MODIFIED, tx);
mutex_exit(&dzp->z_lock);
if (zp->z_zfsvfs->z_norm) {
if (((zp->z_zfsvfs->z_case == ZFS_CASE_INSENSITIVE) &&
(flag & ZCIEXACT)) ||
((zp->z_zfsvfs->z_case == ZFS_CASE_MIXED) &&
!(flag & ZCILOOK)))
error = zap_remove_norm(zp->z_zfsvfs->z_os,
dzp->z_id, dl->dl_name, MT_EXACT, tx);
else
error = zap_remove_norm(zp->z_zfsvfs->z_os,
dzp->z_id, dl->dl_name, MT_FIRST, tx);
} else {
error = zap_remove(zp->z_zfsvfs->z_os,
dzp->z_id, dl->dl_name, tx);
}
ASSERT(error == 0);
if (unlinkedp != NULL)
*unlinkedp = unlinked;
else if (unlinked)
zfs_unlinked_add(zp, tx);
return (0);
}
void
zfs_rmnode(znode_t *zp)
{
zfsvfs_t *zfsvfs = zp->z_zfsvfs;
objset_t *os = zfsvfs->z_os;
znode_t *xzp = NULL;
char obj_name[17];
dmu_tx_t *tx;
uint64_t acl_obj;
int error;
ASSERT(ZTOV(zp)->v_count == 0);
ASSERT(zp->z_phys->zp_links == 0);
/*
* If this is an attribute directory, purge its contents.
*/
if (ZTOV(zp)->v_type == VDIR && (zp->z_phys->zp_flags & ZFS_XATTR))
if (zfs_purgedir(zp) != 0) {
zfs_delete_t *delq = &zfsvfs->z_delete_head;
/*
* Add this back to the delete list to be retried later.
*
* XXX - this could just busy loop on us...
*/
mutex_enter(&delq->z_mutex);
list_insert_tail(&delq->z_znodes, zp);
delq->z_znode_count++;
mutex_exit(&delq->z_mutex);
return;
}
/*
* If the file has extended attributes, unlink the xattr dir.
*/
if (zp->z_phys->zp_xattr) {
error = zfs_zget(zfsvfs, zp->z_phys->zp_xattr, &xzp);
ASSERT(error == 0);
}
acl_obj = zp->z_phys->zp_acl.z_acl_extern_obj;
/*
* Set up the transaction.
*/
tx = dmu_tx_create(os);
dmu_tx_hold_free(tx, zp->z_id, 0, DMU_OBJECT_END);
dmu_tx_hold_zap(tx, zfsvfs->z_dqueue, FALSE, NULL);
if (xzp) {
dmu_tx_hold_bonus(tx, xzp->z_id);
dmu_tx_hold_zap(tx, zfsvfs->z_dqueue, TRUE, NULL);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
zfs_delete_t *delq = &zfsvfs->z_delete_head;
dmu_tx_abort(tx);
/*
* Add this back to the delete list to be retried later.
*
* XXX - this could just busy loop on us...
*/
mutex_enter(&delq->z_mutex);
list_insert_tail(&delq->z_znodes, zp);
delq->z_znode_count++;
mutex_exit(&delq->z_mutex);
return;
}
if (xzp) {
dmu_buf_will_dirty(xzp->z_dbuf, tx);
mutex_enter(&xzp->z_lock);
xzp->z_reap = 1; /* mark xzp for deletion */
xzp->z_phys->zp_links = 0; /* no more links to it */
mutex_exit(&xzp->z_lock);
zfs_dq_add(xzp, tx); /* add xzp to delete queue */
}
/*
* Remove this znode from delete queue
*/
error = zap_remove(os, zfsvfs->z_dqueue,
zfs_dq_hexname(obj_name, zp->z_id), tx);
ASSERT3U(error, ==, 0);
zfs_znode_delete(zp, tx);
dmu_tx_commit(tx);
if (xzp)
VN_RELE(ZTOV(xzp));
}
static
#endif
int
zfs_fuid_find_by_domain(zfsvfs_t *zfsvfs, const char *domain, char **retdomain,
dmu_tx_t *tx)
{
fuid_domain_t searchnode, *findnode;
avl_index_t loc;
/*
* If the dummy "nobody" domain then return an index of 0
* to cause the created FUID to be a standard POSIX id
* for the user nobody.
*/
if (domain[0] == '\0') {
*retdomain = "";
return (0);
}
searchnode.f_ksid = ksid_lookupdomain(domain);
if (retdomain) {
*retdomain = searchnode.f_ksid->kd_name;
}
if (!zfsvfs->z_fuid_loaded)
zfs_fuid_init(zfsvfs, tx);
rw_enter(&zfsvfs->z_fuid_lock, RW_READER);
findnode = avl_find(&zfsvfs->z_fuid_domain, &searchnode, &loc);
rw_exit(&zfsvfs->z_fuid_lock);
if (findnode) {
ksiddomain_rele(searchnode.f_ksid);
return (findnode->f_idx);
} else {
fuid_domain_t *domnode;
nvlist_t *nvp;
nvlist_t **fuids;
uint64_t retidx;
size_t nvsize = 0;
char *packed;
dmu_buf_t *db;
int i = 0;
domnode = kmem_alloc(sizeof (fuid_domain_t), KM_SLEEP);
domnode->f_ksid = searchnode.f_ksid;
rw_enter(&zfsvfs->z_fuid_lock, RW_WRITER);
retidx = domnode->f_idx = avl_numnodes(&zfsvfs->z_fuid_idx) + 1;
avl_add(&zfsvfs->z_fuid_domain, domnode);
avl_add(&zfsvfs->z_fuid_idx, domnode);
/*
* Now resync the on-disk nvlist.
*/
VERIFY(nvlist_alloc(&nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0);
domnode = avl_first(&zfsvfs->z_fuid_domain);
fuids = kmem_alloc(retidx * sizeof (void *), KM_SLEEP);
while (domnode) {
VERIFY(nvlist_alloc(&fuids[i],
NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(fuids[i], FUID_IDX,
domnode->f_idx) == 0);
VERIFY(nvlist_add_uint64(fuids[i],
FUID_OFFSET, 0) == 0);
VERIFY(nvlist_add_string(fuids[i++], FUID_DOMAIN,
domnode->f_ksid->kd_name) == 0);
domnode = AVL_NEXT(&zfsvfs->z_fuid_domain, domnode);
}
VERIFY(nvlist_add_nvlist_array(nvp, FUID_NVP_ARRAY,
fuids, retidx) == 0);
for (i = 0; i != retidx; i++)
nvlist_free(fuids[i]);
kmem_free(fuids, retidx * sizeof (void *));
VERIFY(nvlist_size(nvp, &nvsize, NV_ENCODE_XDR) == 0);
packed = kmem_alloc(nvsize, KM_SLEEP);
VERIFY(nvlist_pack(nvp, &packed, &nvsize,
NV_ENCODE_XDR, KM_SLEEP) == 0);
nvlist_free(nvp);
zfsvfs->z_fuid_size = nvsize;
dmu_write(zfsvfs->z_os, zfsvfs->z_fuid_obj, 0,
zfsvfs->z_fuid_size, packed, tx);
kmem_free(packed, zfsvfs->z_fuid_size);
VERIFY(0 == dmu_bonus_hold(zfsvfs->z_os, zfsvfs->z_fuid_obj,
FTAG, &db));
dmu_buf_will_dirty(db, tx);
*(uint64_t *)db->db_data = zfsvfs->z_fuid_size;
dmu_buf_rele(db, FTAG);
rw_exit(&zfsvfs->z_fuid_lock);
return (retidx);
}
}
/*
* Find layout that corresponds to ordering of attributes
* If not found a new layout number is created and added to
* persistent layout tables.
*/
static int
sa_build_layouts(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc, int attr_count,
dmu_tx_t *tx)
{
sa_os_t *sa = hdl->sa_os->os_sa;
uint64_t hash;
sa_buf_type_t buftype;
sa_hdr_phys_t *sahdr;
void *data_start;
int buf_space;
sa_attr_type_t *attrs, *attrs_start;
int i, lot_count;
int hdrsize, spillhdrsize;
int used;
dmu_object_type_t bonustype;
sa_lot_t *lot;
int len_idx;
int spill_used;
boolean_t spilling;
dmu_buf_will_dirty(hdl->sa_bonus, tx);
bonustype = SA_BONUSTYPE_FROM_DB(hdl->sa_bonus);
/* first determine bonus header size and sum of all attributes */
hdrsize = sa_find_sizes(sa, attr_desc, attr_count, hdl->sa_bonus,
SA_BONUS, &i, &used, &spilling);
if (used > SPA_MAXBLOCKSIZE)
return (EFBIG);
VERIFY(0 == dmu_set_bonus(hdl->sa_bonus, spilling ?
MIN(DN_MAX_BONUSLEN - sizeof (blkptr_t), used + hdrsize) :
used + hdrsize, tx));
ASSERT((bonustype == DMU_OT_ZNODE && spilling == 0) ||
bonustype == DMU_OT_SA);
/* setup and size spill buffer when needed */
if (spilling) {
boolean_t dummy;
if (hdl->sa_spill == NULL) {
VERIFY(dmu_spill_hold_by_bonus(hdl->sa_bonus, NULL,
&hdl->sa_spill) == 0);
}
dmu_buf_will_dirty(hdl->sa_spill, tx);
spillhdrsize = sa_find_sizes(sa, &attr_desc[i],
attr_count - i, hdl->sa_spill, SA_SPILL, &i,
&spill_used, &dummy);
if (spill_used > SPA_MAXBLOCKSIZE)
return (EFBIG);
buf_space = hdl->sa_spill->db_size - spillhdrsize;
if (BUF_SPACE_NEEDED(spill_used, spillhdrsize) >
hdl->sa_spill->db_size)
VERIFY(0 == sa_resize_spill(hdl,
BUF_SPACE_NEEDED(spill_used, spillhdrsize), tx));
}
/* setup starting pointers to lay down data */
data_start = (void *)((uintptr_t)hdl->sa_bonus->db_data + hdrsize);
sahdr = (sa_hdr_phys_t *)hdl->sa_bonus->db_data;
buftype = SA_BONUS;
if (spilling)
buf_space = (sa->sa_force_spill) ?
0 : SA_BLKPTR_SPACE - hdrsize;
else
buf_space = hdl->sa_bonus->db_size - hdrsize;
attrs_start = attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
KM_SLEEP);
lot_count = 0;
for (i = 0, len_idx = 0, hash = -1ULL; i != attr_count; i++) {
uint16_t length;
attrs[i] = attr_desc[i].sa_attr;
length = SA_REGISTERED_LEN(sa, attrs[i]);
if (length == 0)
length = attr_desc[i].sa_length;
if (buf_space < length) { /* switch to spill buffer */
VERIFY(bonustype == DMU_OT_SA);
if (buftype == SA_BONUS && !sa->sa_force_spill) {
sa_find_layout(hdl->sa_os, hash, attrs_start,
lot_count, tx, &lot);
SA_SET_HDR(sahdr, lot->lot_num, hdrsize);
}
buftype = SA_SPILL;
hash = -1ULL;
len_idx = 0;
sahdr = (sa_hdr_phys_t *)hdl->sa_spill->db_data;
sahdr->sa_magic = SA_MAGIC;
data_start = (void *)((uintptr_t)sahdr +
spillhdrsize);
attrs_start = &attrs[i];
buf_space = hdl->sa_spill->db_size - spillhdrsize;
lot_count = 0;
}
hash ^= SA_ATTR_HASH(attrs[i]);
attr_desc[i].sa_addr = data_start;
attr_desc[i].sa_size = length;
SA_COPY_DATA(attr_desc[i].sa_data_func, attr_desc[i].sa_data,
data_start, length);
if (sa->sa_attr_table[attrs[i]].sa_length == 0) {
sahdr->sa_lengths[len_idx++] = length;
}
data_start = (void *)P2ROUNDUP(((uintptr_t)data_start +
length), 8);
buf_space -= P2ROUNDUP(length, 8);
lot_count++;
}
sa_find_layout(hdl->sa_os, hash, attrs_start, lot_count, tx, &lot);
/*
* Verify that old znodes always have layout number 0.
* Must be DMU_OT_SA for arbitrary layouts
*/
VERIFY((bonustype == DMU_OT_ZNODE && lot->lot_num == 0) ||
(bonustype == DMU_OT_SA && lot->lot_num > 1));
if (bonustype == DMU_OT_SA) {
SA_SET_HDR(sahdr, lot->lot_num,
buftype == SA_BONUS ? hdrsize : spillhdrsize);
}
kmem_free(attrs, sizeof (sa_attr_type_t) * attr_count);
if (hdl->sa_bonus_tab) {
sa_idx_tab_rele(hdl->sa_os, hdl->sa_bonus_tab);
hdl->sa_bonus_tab = NULL;
}
if (!sa->sa_force_spill)
VERIFY(0 == sa_build_index(hdl, SA_BONUS));
if (hdl->sa_spill) {
sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
if (!spilling) {
/*
* remove spill block that is no longer needed.
*/
dmu_buf_rele(hdl->sa_spill, NULL);
hdl->sa_spill = NULL;
hdl->sa_spill_tab = NULL;
VERIFY(0 == dmu_rm_spill(hdl->sa_os,
sa_handle_object(hdl), tx));
} else {
VERIFY(0 == sa_build_index(hdl, SA_SPILL));
}
}
return (0);
}
