/*
* Delete the entire contents of a directory. Return a count
* of the number of entries that could not be deleted. If we encounter
* an error, return a count of at least one so that the directory stays
* in the unlinked set.
*
* NOTE: this function assumes that the directory is inactive,
* so there is no need to lock its entries before deletion.
* Also, it assumes the directory contents is *only* regular
* files.
*/
static int
zfs_purgedir(znode_t *dzp)
{
zap_cursor_t zc;
zap_attribute_t zap;
znode_t *xzp;
dmu_tx_t *tx;
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zfs_dirlock_t dl;
int skipped = 0;
int error;
for (zap_cursor_init(&zc, zfsvfs->z_os, dzp->z_id);
(error = zap_cursor_retrieve(&zc, &zap)) == 0;
zap_cursor_advance(&zc)) {
error = zfs_zget(zfsvfs,
ZFS_DIRENT_OBJ(zap.za_first_integer), &xzp);
if (error) {
skipped += 1;
continue;
}
/*
ASSERT((ZTOV(xzp)->v_type == VREG) ||
(ZTOV(xzp)->v_type == VLNK));
*/
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, zap.za_name);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/* Is this really needed ? */
zfs_sa_upgrade_txholds(tx, xzp);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
//VN_RELE(ZTOV(xzp)); // async
VN_RELE_ASYNC(ZTOV(xzp), dsl_pool_vnrele_taskq(dmu_objset_pool(zfsvfs->z_os)));
skipped += 1;
continue;
}
bzero(&dl, sizeof (dl));
dl.dl_dzp = dzp;
dl.dl_name = zap.za_name;
error = zfs_link_destroy(&dl, xzp, tx, 0, NULL);
if (error)
skipped += 1;
dmu_tx_commit(tx);
//VN_RELE(ZTOV(xzp)); // async
VN_RELE_ASYNC(ZTOV(xzp), dsl_pool_vnrele_taskq(dmu_objset_pool(zfsvfs->z_os)));
}
zap_cursor_fini(&zc);
if (error != ENOENT)
skipped += 1;
return (skipped);
}
/*
* Delete the entire contents of a directory. Return a count
* of the number of entries that could not be deleted. If we encounter
* an error, return a count of at least one so that the directory stays
* in the unlinked set.
*
* NOTE: this function assumes that the directory is inactive,
* so there is no need to lock its entries before deletion.
* Also, it assumes the directory contents is *only* regular
* files.
*/
static int
zfs_purgedir(znode_t *dzp)
{
zap_cursor_t zc;
zap_attribute_t zap;
znode_t *xzp;
dmu_tx_t *tx;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zfs_dirlock_t dl;
int skipped = 0;
int error;
for (zap_cursor_init(&zc, zfsvfs->z_os, dzp->z_id);
(error = zap_cursor_retrieve(&zc, &zap)) == 0;
zap_cursor_advance(&zc)) {
error = zfs_zget(zfsvfs,
ZFS_DIRENT_OBJ(zap.za_first_integer), &xzp);
if (error) {
skipped += 1;
continue;
}
ASSERT(S_ISREG(ZTOI(xzp)->i_mode) ||
S_ISLNK(ZTOI(xzp)->i_mode));
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, zap.za_name);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/* Is this really needed ? */
zfs_sa_upgrade_txholds(tx, xzp);
dmu_tx_mark_netfree(tx);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
zfs_iput_async(ZTOI(xzp));
skipped += 1;
continue;
}
bzero(&dl, sizeof (dl));
dl.dl_dzp = dzp;
dl.dl_name = zap.za_name;
error = zfs_link_destroy(&dl, xzp, tx, 0, NULL);
if (error)
skipped += 1;
dmu_tx_commit(tx);
zfs_iput_async(ZTOI(xzp));
}
zap_cursor_fini(&zc);
if (error != ENOENT)
skipped += 1;
return (skipped);
}
/*
* Delete the entire contents of a directory. Return a count
* of the number of entries that could not be deleted. If we encounter
* an error, return a count of at least one so that the directory stays
* in the unlinked set.
*
* NOTE: this function assumes that the directory is inactive,
* so there is no need to lock its entries before deletion.
* Also, it assumes the directory contents is *only* regular
* files.
*/
static int
zfs_purgedir(znode_t *dzp)
{
zap_cursor_t zc;
zap_attribute_t zap;
znode_t *xzp;
dmu_tx_t *tx;
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
int skipped = 0;
int error;
for (zap_cursor_init(&zc, zfsvfs->z_os, dzp->z_id);
(error = zap_cursor_retrieve(&zc, &zap)) == 0;
zap_cursor_advance(&zc)) {
error = zfs_zget(zfsvfs,
ZFS_DIRENT_OBJ(zap.za_first_integer), &xzp);
if (error) {
skipped += 1;
continue;
}
vn_lock(ZTOV(xzp), LK_EXCLUSIVE | LK_RETRY);
ASSERT((ZTOV(xzp)->v_type == VREG) ||
(ZTOV(xzp)->v_type == VLNK));
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, zap.za_name);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/* Is this really needed ? */
zfs_sa_upgrade_txholds(tx, xzp);
dmu_tx_mark_netfree(tx);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
vput(ZTOV(xzp));
skipped += 1;
continue;
}
error = zfs_link_destroy(dzp, zap.za_name, xzp, tx, 0, NULL);
if (error)
skipped += 1;
dmu_tx_commit(tx);
vput(ZTOV(xzp));
}
zap_cursor_fini(&zc);
if (error != ENOENT)
skipped += 1;
return (skipped);
}
static void
__osd_xattr_declare_del(const struct lu_env *env, struct osd_object *obj,
const char *name, struct osd_thandle *oh)
{
struct osd_device *osd = osd_obj2dev(obj);
dmu_tx_t *tx = oh->ot_tx;
uint64_t xa_data_obj;
int rc;
/* update SA_ZPL_DXATTR if xattr was in SA */
dmu_tx_hold_sa(tx, obj->oo_sa_hdl, 0);
if (obj->oo_xattr == ZFS_NO_OBJECT)
return;
rc = -zap_lookup(osd->od_os, obj->oo_xattr, name, 8, 1, &xa_data_obj);
if (rc == 0) {
/*
* Entry exists.
* We'll delete the existing object and ZAP entry.
*/
dmu_tx_hold_bonus(tx, xa_data_obj);
dmu_tx_hold_free(tx, xa_data_obj, 0, DMU_OBJECT_END);
dmu_tx_hold_zap(tx, obj->oo_xattr, FALSE, (char *) name);
return;
} else if (rc == -ENOENT) {
/*
* Entry doesn't exist, nothing to be changed.
*/
return;
}
/* An error happened */
tx->tx_err = -rc;
}
void
zfs_sa_upgrade_txholds(dmu_tx_t *tx, znode_t *zp)
{
if (!zp->z_zfsvfs->z_use_sa || zp->z_is_sa)
return;
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
if (zfs_external_acl(zp)) {
dmu_tx_hold_free(tx, zfs_external_acl(zp), 0,
DMU_OBJECT_END);
}
}
/*
* TX_WRITE2 are only generated when dmu_sync() returns EALREADY
* meaning the pool block is already being synced. So now that we always write
* out full blocks, all we have to do is expand the eof if
* the file is grown.
*/
static int
zfs_replay_write2(zfsvfs_t *zfsvfs, void *data, boolean_t byteswap)
{
#ifndef TODO_OSV
kprintf("TX_WRITE2\n");
return EOPNOTSUPP;
#else
lr_write_t *lr = data;
znode_t *zp;
int error;
uint64_t end;
if (byteswap)
byteswap_uint64_array(lr, sizeof (*lr));
if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0)
return (error);
top:
end = lr->lr_offset + lr->lr_length;
if (end > zp->z_size) {
dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);
zp->z_size = end;
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
VN_RELE(ZTOV(zp));
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
return (error);
}
(void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
(void *)&zp->z_size, sizeof (uint64_t), tx);
/* Ensure the replayed seq is updated */
(void) zil_replaying(zfsvfs->z_log, tx);
dmu_tx_commit(tx);
}
VN_RELE(ZTOV(zp));
return (error);
#endif
}
/*
* TX_WRITE2 are only generated when dmu_sync() returns EALREADY
* meaning the pool block is already being synced. So now that we always write
* out full blocks, all we have to do is expand the eof if
* the file is grown.
*/
static int
zfs_replay_write2(void *arg1, void *arg2, boolean_t byteswap)
{
zfsvfs_t *zfsvfs = arg1;
lr_write_t *lr = arg2;
znode_t *zp;
int error;
uint64_t end;
if (byteswap)
byteswap_uint64_array(lr, sizeof (*lr));
if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0)
return (error);
top:
end = lr->lr_offset + lr->lr_length;
if (end > zp->z_size) {
dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);
zp->z_size = end;
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
iput(ZTOI(zp));
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
return (error);
}
(void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
(void *)&zp->z_size, sizeof (uint64_t), tx);
/* Ensure the replayed seq is updated */
(void) zil_replaying(zfsvfs->z_log, tx);
dmu_tx_commit(tx);
}
iput(ZTOI(zp));
return (error);
}
static ssize_t osd_declare_write(const struct lu_env *env, struct dt_object *dt,
const struct lu_buf *buf, loff_t pos,
struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t oid;
ENTRY;
oh = container_of0(th, struct osd_thandle, ot_super);
/* in some cases declare can race with creation (e.g. llog)
* and we need to wait till object is initialized. notice
* LOHA_EXISTs is supposed to be the last step in the
* initialization */
/* declare possible size change. notice we can't check
* current size here as another thread can change it */
if (dt_object_exists(dt)) {
LASSERT(obj->oo_db);
oid = obj->oo_db->db_object;
dmu_tx_hold_sa(oh->ot_tx, obj->oo_sa_hdl, 0);
} else {
oid = DMU_NEW_OBJECT;
dmu_tx_hold_sa_create(oh->ot_tx, ZFS_SA_BASE_ATTR_SIZE);
}
/* XXX: we still miss for append declaration support in ZFS
* -1 means append which is used by llog mostly, llog
* can grow upto LLOG_MIN_CHUNK_SIZE*8 records */
if (pos == -1)
pos = max_t(loff_t, 256 * 8 * LLOG_MIN_CHUNK_SIZE,
obj->oo_attr.la_size + (2 << 20));
dmu_tx_hold_write(oh->ot_tx, oid, pos, buf->lb_len);
/* dt_declare_write() is usually called for system objects, such
* as llog or last_rcvd files. We needn't enforce quota on those
* objects, so always set the lqi_space as 0. */
RETURN(osd_declare_quota(env, osd, obj->oo_attr.la_uid,
obj->oo_attr.la_gid, 0, oh, true, NULL,
false));
}
int
zfs_sa_set_xattr(znode_t *zp)
{
zfs_sb_t *zsb = ZTOZSB(zp);
dmu_tx_t *tx;
char *obj;
size_t size;
int error;
ASSERT(RW_WRITE_HELD(&zp->z_xattr_lock));
ASSERT(zp->z_xattr_cached);
ASSERT(zp->z_is_sa);
error = nvlist_size(zp->z_xattr_cached, &size, NV_ENCODE_XDR);
if (error)
goto out;
obj = zio_buf_alloc(size);
error = nvlist_pack(zp->z_xattr_cached, &obj, &size,
NV_ENCODE_XDR, KM_SLEEP);
if (error)
goto out_free;
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_sa_create(tx, size);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
} else {
error = sa_update(zp->z_sa_hdl, SA_ZPL_DXATTR(zsb),
obj, size, tx);
if (error)
dmu_tx_abort(tx);
else
dmu_tx_commit(tx);
}
out_free:
zio_buf_free(obj, size);
out:
return (error);
}
int
zfs_sa_set_xattr(znode_t *zp)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_tx_t *tx;
char *obj;
size_t size;
int error;
ASSERT(RW_WRITE_HELD(&zp->z_xattr_lock));
ASSERT(zp->z_xattr_cached);
ASSERT(zp->z_is_sa);
error = nvlist_size(zp->z_xattr_cached, &size, NV_ENCODE_XDR);
if ((error == 0) && (size > SA_ATTR_MAX_LEN))
error = EFBIG;
if (error)
goto out;
obj = vmem_alloc(size, KM_SLEEP);
error = nvlist_pack(zp->z_xattr_cached, &obj, &size,
NV_ENCODE_XDR, KM_SLEEP);
if (error)
goto out_free;
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa_create(tx, size);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
} else {
VERIFY0(sa_update(zp->z_sa_hdl, SA_ZPL_DXATTR(zfsvfs),
obj, size, tx));
dmu_tx_commit(tx);
}
out_free:
vmem_free(obj, size);
out:
return (error);
}
static ssize_t osd_declare_write(const struct lu_env *env, struct dt_object *dt,
const loff_t size, loff_t pos,
struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t oid;
ENTRY;
oh = container_of0(th, struct osd_thandle, ot_super);
/* in some cases declare can race with creation (e.g. llog)
* and we need to wait till object is initialized. notice
* LOHA_EXISTs is supposed to be the last step in the
* initialization */
/* declare possible size change. notice we can't check
* current size here as another thread can change it */
if (dt_object_exists(dt)) {
LASSERT(obj->oo_db);
oid = obj->oo_db->db_object;
dmu_tx_hold_sa(oh->ot_tx, obj->oo_sa_hdl, 0);
} else {
oid = DMU_NEW_OBJECT;
dmu_tx_hold_sa_create(oh->ot_tx, ZFS_SA_BASE_ATTR_SIZE);
}
dmu_tx_hold_write(oh->ot_tx, oid, pos, size);
/* dt_declare_write() is usually called for system objects, such
* as llog or last_rcvd files. We needn't enforce quota on those
* objects, so always set the lqi_space as 0. */
RETURN(osd_declare_quota(env, osd, obj->oo_attr.la_uid,
obj->oo_attr.la_gid, 0, oh, true, NULL,
false));
}
int
zfs_make_xattrdir(znode_t *zp, vattr_t *vap, struct inode **xipp, cred_t *cr)
{
zfs_sb_t *zsb = ZTOZSB(zp);
znode_t *xzp;
dmu_tx_t *tx;
int error;
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
#ifdef DEBUG
uint64_t parent;
#endif
*xipp = NULL;
if ((error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, 0, B_FALSE, cr)))
return (error);
if ((error = zfs_acl_ids_create(zp, IS_XATTR, vap, cr, NULL,
&acl_ids)) != 0)
return (error);
if (zfs_acl_ids_overquota(zsb, &acl_ids)) {
zfs_acl_ids_free(&acl_ids);
return (SET_ERROR(EDQUOT));
}
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
fuid_dirtied = zsb->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zsb, tx);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, tx, cr, IS_XATTR, &xzp, &acl_ids);
if (fuid_dirtied)
zfs_fuid_sync(zsb, tx);
#ifdef DEBUG
error = sa_lookup(xzp->z_sa_hdl, SA_ZPL_PARENT(zsb),
&parent, sizeof (parent));
ASSERT(error == 0 && parent == zp->z_id);
#endif
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zsb), &xzp->z_id,
sizeof (xzp->z_id), tx));
if (!zp->z_unlinked)
(void) zfs_log_create(zsb->z_log, tx, TX_MKXATTR, zp,
xzp, "", NULL, acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
*xipp = ZTOI(xzp);
return (0);
}
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;
uint64_t xattr_obj;
int error;
ASSERT(zp->z_links == 0);
ASSERT(ZTOV(zp)->v_count == 0);
/*
* If this is an attribute directory, purge its contents.
*/
if (ZTOV(zp)->v_type == VDIR && (zp->z_pflags & 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.
*/
error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (error == 0 && xattr_obj) {
error = zfs_zget(zfsvfs, xattr_obj, &xzp);
ASSERT(error == 0);
}
acl_obj = zfs_external_acl(zp);
/*
* 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_zap(tx, zfsvfs->z_unlinkedobj, TRUE, NULL);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
zfs_sa_upgrade_txholds(tx, zp);
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) {
ASSERT(error == 0);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE; /* mark xzp for deletion */
xzp->z_links = 0; /* no more links to it */
VERIFY(0 == sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
&xzp->z_links, sizeof (xzp->z_links), tx));
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));
}
/*
* Link zp into dl. Can only fail if zp has been unlinked.
*/
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_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
uint64_t parent;
*xvpp = NULL;
if ((error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, 0, B_FALSE, cr)))
return (error);
if ((error = zfs_acl_ids_create(zp, IS_XATTR, vap, cr, NULL,
&acl_ids)) != 0)
return (error);
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
zfs_acl_ids_free(&acl_ids);
return (EDQUOT);
}
top:
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, tx, cr, IS_XATTR, &xzp, &acl_ids);
#ifdef HAVE_ZPL
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
#endif /* HAVE_ZPL */
#ifdef DEBUG
error = sa_lookup(xzp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (parent));
ASSERT(error == 0 && parent == zp->z_id);
#endif
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xzp->z_id,
sizeof (xzp->z_id), tx));
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp,
xzp, "", NULL, acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
*xvpp = ZTOV(xzp);
return (0);
}
static int osd_declare_attr_set(const struct lu_env *env,
struct dt_object *dt,
const struct lu_attr *attr,
struct thandle *handle)
{
struct osd_thread_info *info = osd_oti_get(env);
char *buf = osd_oti_get(env)->oti_str;
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t bspace;
uint32_t blksize;
int rc = 0;
ENTRY;
LASSERT(handle != NULL);
LASSERT(osd_invariant(obj));
oh = container_of0(handle, struct osd_thandle, ot_super);
down_read(&obj->oo_guard);
if (unlikely(!dt_object_exists(dt) || obj->oo_destroyed))
GOTO(out, rc = 0);
LASSERT(obj->oo_sa_hdl != NULL);
LASSERT(oh->ot_tx != NULL);
dmu_tx_hold_sa(oh->ot_tx, obj->oo_sa_hdl, 0);
if (oh->ot_tx->tx_err != 0)
GOTO(out, rc = -oh->ot_tx->tx_err);
sa_object_size(obj->oo_sa_hdl, &blksize, &bspace);
bspace = toqb(bspace * blksize);
__osd_xattr_declare_set(env, obj, sizeof(struct lustre_mdt_attrs),
XATTR_NAME_LMA, oh);
if (attr && attr->la_valid & LA_UID) {
/* account for user inode tracking ZAP update */
dmu_tx_hold_bonus(oh->ot_tx, osd->od_iusr_oid);
dmu_tx_hold_zap(oh->ot_tx, osd->od_iusr_oid, TRUE, buf);
/* quota enforcement for user */
if (attr->la_uid != obj->oo_attr.la_uid) {
rc = qsd_transfer(env, osd->od_quota_slave,
&oh->ot_quota_trans, USRQUOTA,
obj->oo_attr.la_uid, attr->la_uid,
bspace, &info->oti_qi);
if (rc)
GOTO(out, rc);
}
}
if (attr && attr->la_valid & LA_GID) {
/* account for user inode tracking ZAP update */
dmu_tx_hold_bonus(oh->ot_tx, osd->od_igrp_oid);
dmu_tx_hold_zap(oh->ot_tx, osd->od_igrp_oid, TRUE, buf);
/* quota enforcement for group */
if (attr->la_gid != obj->oo_attr.la_gid) {
rc = qsd_transfer(env, osd->od_quota_slave,
&oh->ot_quota_trans, GRPQUOTA,
obj->oo_attr.la_gid, attr->la_gid,
bspace, &info->oti_qi);
if (rc)
GOTO(out, rc);
}
}
out:
up_read(&obj->oo_guard);
RETURN(rc);
}
void __osd_xattr_declare_set(const struct lu_env *env, struct osd_object *obj,
int vallen, const char *name,
struct osd_thandle *oh)
{
struct osd_device *osd = osd_obj2dev(obj);
dmu_buf_t *db = obj->oo_db;
dmu_tx_t *tx = oh->ot_tx;
uint64_t xa_data_obj;
int rc = 0;
int here;
if (unlikely(obj->oo_destroyed))
return;
here = dt_object_exists(&obj->oo_dt);
/* object may be not yet created */
if (here) {
LASSERT(db);
LASSERT(obj->oo_sa_hdl);
/* we might just update SA_ZPL_DXATTR */
dmu_tx_hold_sa(tx, obj->oo_sa_hdl, 1);
if (obj->oo_xattr == ZFS_NO_OBJECT)
rc = -ENOENT;
}
if (!here || rc == -ENOENT) {
/* we'll be updating SA_ZPL_XATTR */
if (here) {
LASSERT(obj->oo_sa_hdl);
dmu_tx_hold_sa(tx, obj->oo_sa_hdl, 1);
}
/* xattr zap + entry */
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, TRUE, (char *) name);
/* xattr value obj */
dmu_tx_hold_sa_create(tx, ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, vallen);
return;
}
rc = -zap_lookup(osd->od_os, obj->oo_xattr, name, sizeof(uint64_t), 1,
&xa_data_obj);
if (rc == 0) {
/*
* Entry already exists.
* We'll truncate the existing object.
*/
dmu_tx_hold_bonus(tx, xa_data_obj);
dmu_tx_hold_free(tx, xa_data_obj, vallen, DMU_OBJECT_END);
dmu_tx_hold_write(tx, xa_data_obj, 0, vallen);
return;
} else if (rc == -ENOENT) {
/*
* Entry doesn't exist, we need to create a new one and a new
* object to store the value.
*/
dmu_tx_hold_bonus(tx, obj->oo_xattr);
dmu_tx_hold_zap(tx, obj->oo_xattr, TRUE, (char *) name);
dmu_tx_hold_sa_create(tx, ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, vallen);
return;
}
/* An error happened */
tx->tx_err = -rc;
}
/*
* Lookup/Create an extended attribute entry.
*
* Input arguments:
* dzp - znode for hidden attribute directory
* name - name of attribute
* flag - ZNEW: if the entry already exists, fail with EEXIST.
* ZEXISTS: if the entry does not exist, fail with ENOENT.
*
* Output arguments:
* vpp - pointer to the vnode for the entry (NULL if there isn't one)
*
* Return value: 0 on success or errno value on failure.
*/
int
zfs_obtain_xattr(znode_t *dzp, const char *name, mode_t mode, cred_t *cr,
vnode_t **vpp, int flag)
{
znode_t *xzp = NULL;
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
struct vnode_attr vattr;
int error;
struct componentname cn;
zfs_acl_ids_t acl_ids;
/* zfs_dirent_lock() expects a component name */
bzero(&cn, sizeof (cn));
cn.cn_nameiop = LOOKUP;
cn.cn_flags = ISLASTCN;
cn.cn_nameptr = (char *)name;
cn.cn_namelen = strlen(name);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
zilog = zfsvfs->z_log;
VATTR_INIT(&vattr);
VATTR_SET(&vattr, va_type, VREG);
VATTR_SET(&vattr, va_mode, mode & ~S_IFMT);
if ((error = zfs_acl_ids_create(dzp, 0,
&vattr, cr, NULL, &acl_ids)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
top:
/* Lock the attribute entry name. */
if ( (error = zfs_dirent_lock(&dl, dzp, (char *)name, &xzp, flag,
NULL, &cn)) ) {
goto out;
}
/* If the name already exists, we're done. */
if (xzp != NULL) {
zfs_dirent_unlock(dl);
goto out;
}
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
//dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
//dmu_tx_hold_bonus(tx, dzp->z_id);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, (char *)name);
#if 1 // FIXME
if (dzp->z_pflags & ZFS_INHERIT_ACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, SPA_MAXBLOCKSIZE);
}
#endif
zfs_sa_upgrade_txholds(tx, dzp);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
goto out;
}
zfs_mknode(dzp, &vattr, tx, cr, 0, &xzp, &acl_ids);
/*
ASSERT(xzp->z_id == zoid);
*/
(void) zfs_link_create(dl, xzp, tx, ZNEW);
zfs_log_create(zilog, tx, TX_CREATE, dzp, xzp, (char *)name,
NULL /* vsecp */, 0 /*acl_ids.z_fuidp*/, &vattr);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
zfs_znode_wait_vnode(xzp);
zfs_dirent_unlock(dl);
out:
if (error == EEXIST)
error = ENOATTR;
if (xzp)
*vpp = ZTOV(xzp);
ZFS_EXIT(zfsvfs);
return (error);
}
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_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
uint64_t parent;
*xvpp = NULL;
/*
* In FreeBSD, access checking for creating an EA is being done
* in zfs_setextattr(),
*/
#ifndef __FreeBSD_kernel__
if (error = zfs_zaccess(zp, ACE_WRITE_NAMED_ATTRS, 0, B_FALSE, cr))
return (error);
#endif
if ((error = zfs_acl_ids_create(zp, IS_XATTR, vap, cr, NULL,
&acl_ids)) != 0)
return (error);
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
zfs_acl_ids_free(&acl_ids);
return (SET_ERROR(EDQUOT));
}
getnewvnode_reserve(1);
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, tx, cr, IS_XATTR, &xzp, &acl_ids);
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
#ifdef DEBUG
error = sa_lookup(xzp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (parent));
ASSERT(error == 0 && parent == zp->z_id);
#endif
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xzp->z_id,
sizeof (xzp->z_id), tx));
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp,
xzp, "", NULL, acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
getnewvnode_drop_reserve();
*xvpp = ZTOV(xzp);
return (0);
}
void
zfs_rmnode(znode_t *zp)
{
zfs_sb_t *zsb = ZTOZSB(zp);
objset_t *os = zsb->z_os;
znode_t *xzp = NULL;
dmu_tx_t *tx;
uint64_t acl_obj;
uint64_t xattr_obj;
uint64_t count;
int error;
ASSERT(zp->z_links == 0);
ASSERT(atomic_read(&ZTOI(zp)->i_count) == 0);
/*
* If this is an attribute directory, purge its contents.
*/
if (S_ISDIR(ZTOI(zp)->i_mode) && (zp->z_pflags & ZFS_XATTR)) {
error = zap_count(os, zp->z_id, &count);
if (error) {
zfs_znode_dmu_fini(zp);
return;
}
if (count > 0) {
taskq_t *taskq;
/*
* There are still directory entries in this xattr
* directory. Let zfs_unlinked_drain() deal with
* them to avoid deadlocking this process in the
* zfs_purgedir()->zfs_zget()->ilookup() callpath
* on the xattr inode's I_FREEING bit.
*/
taskq = dsl_pool_iput_taskq(dmu_objset_pool(os));
taskq_dispatch(taskq, (task_func_t *)
zfs_unlinked_drain, zsb, TQ_SLEEP);
zfs_znode_dmu_fini(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);
return;
}
/*
* If the file has extended attributes, we're going to unlink
* the xattr dir.
*/
error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zsb),
&xattr_obj, sizeof (xattr_obj));
if (error == 0 && xattr_obj) {
error = zfs_zget(zsb, xattr_obj, &xzp);
ASSERT(error == 0);
}
acl_obj = zfs_external_acl(zp);
/*
* 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, zsb->z_unlinkedobj, FALSE, NULL);
if (xzp) {
dmu_tx_hold_zap(tx, zsb->z_unlinkedobj, TRUE, NULL);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
zfs_sa_upgrade_txholds(tx, zp);
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);
goto out;
}
if (xzp) {
ASSERT(error == 0);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE; /* mark xzp for deletion */
xzp->z_links = 0; /* no more links to it */
VERIFY(0 == sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zsb),
&xzp->z_links, sizeof (xzp->z_links), tx));
mutex_exit(&xzp->z_lock);
zfs_unlinked_add(xzp, tx);
}
/* Remove this znode from the unlinked set */
VERIFY3U(0, ==,
zap_remove_int(zsb->z_os, zsb->z_unlinkedobj, zp->z_id, tx));
zfs_znode_delete(zp, tx);
dmu_tx_commit(tx);
out:
if (xzp)
iput(ZTOI(xzp));
}
void
zfs_rmnode(znode_t *zp)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
objset_t *os = zfsvfs->z_os;
znode_t *xzp = NULL;
dmu_tx_t *tx;
uint64_t acl_obj;
uint64_t xattr_obj;
uint64_t links;
int error;
ASSERT(ZTOI(zp)->i_nlink == 0);
ASSERT(atomic_read(&ZTOI(zp)->i_count) == 0);
/*
* If this is an attribute directory, purge its contents.
*/
if (S_ISDIR(ZTOI(zp)->i_mode) && (zp->z_pflags & 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);
return;
}
}
/*
* Free up all the data in the file. We don't do this for directories
* because we need truncate and remove to be in the same tx, like in
* zfs_znode_delete(). Otherwise, if we crash here we'll end up with
* an inconsistent truncated zap object in the delete queue. Note a
* truncated file is harmless since it only contains user data.
*/
if (S_ISREG(ZTOI(zp)->i_mode)) {
error = dmu_free_long_range(os, zp->z_id, 0, DMU_OBJECT_END);
if (error) {
/*
* Not enough space or we were interrupted by unmount.
* Leave the file in the unlinked set.
*/
zfs_znode_dmu_fini(zp);
return;
}
}
/*
* If the file has extended attributes, we're going to unlink
* the xattr dir.
*/
error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (error == 0 && xattr_obj) {
error = zfs_zget(zfsvfs, xattr_obj, &xzp);
ASSERT(error == 0);
}
acl_obj = zfs_external_acl(zp);
/*
* 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_zap(tx, zfsvfs->z_unlinkedobj, TRUE, NULL);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
}
if (acl_obj)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
zfs_sa_upgrade_txholds(tx, zp);
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);
goto out;
}
if (xzp) {
ASSERT(error == 0);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE; /* mark xzp for deletion */
clear_nlink(ZTOI(xzp)); /* no more links to it */
links = 0;
VERIFY(0 == sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
&links, sizeof (links), tx));
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)
zfs_iput_async(ZTOI(xzp));
}
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;
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
*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__*/
if ((error = zfs_acl_ids_create(zp, IS_XATTR, vap, cr, NULL,
&acl_ids)) != 0)
return (error);
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids)) {
zfs_acl_ids_free(&acl_ids);
return (EDQUOT);
}
top:
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
return (error);
}
zfs_mknode(zp, vap, tx, cr, IS_XATTR, &xzp, &acl_ids);
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs), &xzp->z_id,
sizeof (xzp->z_id), tx));
(void) zfs_log_create(zfsvfs->z_log, tx, TX_MKXATTR, zp,
xzp, "", NULL, acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
/* Cleanup any znode we consumed during zfs_mknode() */
printf("ZFS_POSTPROCESS_ZP(xzp);\n");
printf("zfs_dir attach 2\n");
zfs_attach_vnode(xzp);
*xvpp = ZTOV(xzp);
return (0);
}
static int osd_declare_attr_set(const struct lu_env *env,
struct dt_object *dt,
const struct lu_attr *attr,
struct thandle *handle)
{
struct osd_thread_info *info = osd_oti_get(env);
char *buf = osd_oti_get(env)->oti_str;
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t bspace;
uint32_t blksize;
int rc;
ENTRY;
if (!dt_object_exists(dt)) {
/* XXX: sanity check that object creation is declared */
RETURN(0);
}
LASSERT(handle != NULL);
LASSERT(osd_invariant(obj));
oh = container_of0(handle, struct osd_thandle, ot_super);
LASSERT(obj->oo_sa_hdl != NULL);
dmu_tx_hold_sa(oh->ot_tx, obj->oo_sa_hdl, 0);
sa_object_size(obj->oo_sa_hdl, &blksize, &bspace);
bspace = toqb(bspace * blksize);
if (attr && attr->la_valid & LA_UID) {
/* account for user inode tracking ZAP update */
dmu_tx_hold_bonus(oh->ot_tx, osd->od_iusr_oid);
dmu_tx_hold_zap(oh->ot_tx, osd->od_iusr_oid, TRUE, buf);
/* quota enforcement for user */
if (attr->la_uid != obj->oo_attr.la_uid) {
rc = qsd_transfer(env, osd->od_quota_slave,
&oh->ot_quota_trans, USRQUOTA,
obj->oo_attr.la_uid, attr->la_uid,
bspace, &info->oti_qi);
if (rc)
RETURN(rc);
}
}
if (attr && attr->la_valid & LA_GID) {
/* account for user inode tracking ZAP update */
dmu_tx_hold_bonus(oh->ot_tx, osd->od_igrp_oid);
dmu_tx_hold_zap(oh->ot_tx, osd->od_igrp_oid, TRUE, buf);
/* quota enforcement for group */
if (attr->la_gid != obj->oo_attr.la_gid) {
rc = qsd_transfer(env, osd->od_quota_slave,
&oh->ot_quota_trans, GRPQUOTA,
obj->oo_attr.la_gid, attr->la_gid,
bspace, &info->oti_qi);
if (rc)
RETURN(rc);
}
}
RETURN(0);
}
/*
* Delete the entire contents of a directory. Return a count
* of the number of entries that could not be deleted.
*
* NOTE: this function assumes that the directory is inactive,
* so there is no need to lock its entries before deletion.
* Also, it assumes the directory contents is *only* regular
* files.
*/
static int
zfs_purgedir(znode_t *dzp)
{
zap_cursor_t zc;
zap_attribute_t zap;
znode_t *xzp;
dmu_tx_t *tx;
zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
zfs_dirlock_t dl;
int skipped = 0;
int error;
for (zap_cursor_init(&zc, zfsvfs->z_os, dzp->z_id);
(error = zap_cursor_retrieve(&zc, &zap)) == 0;
zap_cursor_advance(&zc)) {
#ifdef __APPLE__
error = zfs_zget_sans_vnode(zfsvfs, ZFS_DIRENT_OBJ(zap.za_first_integer), &xzp);
ASSERT3U(error, ==, 0);
#else
error = zfs_zget(zfsvfs,
ZFS_DIRENT_OBJ(zap.za_first_integer), &xzp);
ASSERT3U(error, ==, 0);
ASSERT((ZTOV(xzp)->v_type == VREG) ||
(ZTOV(xzp)->v_type == VLNK));
#endif /* __APPLE__ */
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, zap.za_name);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/* Is this really needed ? */
zfs_sa_upgrade_txholds(tx, xzp);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
#ifdef __APPLE__
if (ZTOV(xzp) == NULL) {
zfs_zinactive(xzp);
} else {
VN_RELE(ZTOV(xzp));
}
#else
VN_RELE(ZTOV(xzp));
#endif /* __APPLE__ */
skipped += 1;
continue;
}
bzero(&dl, sizeof (dl));
dl.dl_dzp = dzp;
dl.dl_name = zap.za_name;
error = zfs_link_destroy(&dl, xzp, tx, 0, NULL);
ASSERT3U(error, ==, 0);
dmu_tx_commit(tx);
#ifdef __APPLE__
if (ZTOV(xzp) == NULL) {
zfs_zinactive(xzp);
} else {
VN_RELE(ZTOV(xzp));
}
#else
VN_RELE(ZTOV(xzp));
#endif /* __APPLE__ */
}
zap_cursor_fini(&zc);
ASSERT(error == ENOENT);
return (skipped);
}
/*
* Increase the file length
*
* IN: zp - znode of file to free data in.
* end - new end-of-file
*
* RETURN: 0 on success, error code on failure
*/
static int
zfs_extend(znode_t *zp, uint64_t end)
{
zfs_sb_t *zsb = ZTOZSB(zp);
dmu_tx_t *tx;
rl_t *rl;
uint64_t newblksz;
int error;
/*
* We will change zp_size, lock the whole file.
*/
rl = zfs_range_lock(zp, 0, UINT64_MAX, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (end <= zp->z_size) {
zfs_range_unlock(rl);
return (0);
}
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
if (end > zp->z_blksz &&
(!ISP2(zp->z_blksz) || zp->z_blksz < zsb->z_max_blksz)) {
/*
* We are growing the file past the current block size.
*/
if (zp->z_blksz > ZTOZSB(zp)->z_max_blksz) {
ASSERT(!ISP2(zp->z_blksz));
newblksz = MIN(end, SPA_MAXBLOCKSIZE);
} else {
newblksz = MIN(end, ZTOZSB(zp)->z_max_blksz);
}
dmu_tx_hold_write(tx, zp->z_id, 0, newblksz);
} else {
newblksz = 0;
}
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
zfs_range_unlock(rl);
return (error);
}
if (newblksz)
zfs_grow_blocksize(zp, newblksz, tx);
zp->z_size = end;
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(ZTOZSB(zp)),
&zp->z_size, sizeof (zp->z_size), tx));
zfs_range_unlock(rl);
dmu_tx_commit(tx);
return (0);
}
/*
* Free space in a file
*
* IN: zp - znode of file to free data in.
* off - start of range
* len - end of range (0 => EOF)
* flag - current file open mode flags.
* log - TRUE if this action should be logged
*
* RETURN: 0 on success, error code on failure
*/
int
zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log)
{
struct inode *ip = ZTOI(zp);
dmu_tx_t *tx;
zfs_sb_t *zsb = ZTOZSB(zp);
zilog_t *zilog = zsb->z_log;
uint64_t mode;
uint64_t mtime[2], ctime[2];
sa_bulk_attr_t bulk[3];
int count = 0;
int error;
if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zsb), &mode,
sizeof (mode))) != 0)
return (error);
if (off > zp->z_size) {
error = zfs_extend(zp, off+len);
if (error == 0 && log)
goto log;
else
return (error);
}
/*
* Check for any locks in the region to be freed.
*/
if (ip->i_flock && mandatory_lock(ip)) {
uint64_t length = (len ? len : zp->z_size - off);
if (!lock_may_write(ip, off, length))
return (SET_ERROR(EAGAIN));
}
if (len == 0) {
error = zfs_trunc(zp, off);
} else {
if ((error = zfs_free_range(zp, off, len)) == 0 &&
off + len > zp->z_size)
error = zfs_extend(zp, off+len);
}
if (error || !log)
return (error);
log:
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
return (error);
}
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL, mtime, 16);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL, ctime, 16);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb),
NULL, &zp->z_pflags, 8);
zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE);
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
zfs_log_truncate(zilog, tx, TX_TRUNCATE, zp, off, len);
dmu_tx_commit(tx);
zfs_inode_update(zp);
return (0);
}
/*
* Truncate a file
*
* IN: zp - znode of file to free data in.
* end - new end-of-file.
*
* RETURN: 0 on success, error code on failure
*/
static int
zfs_trunc(znode_t *zp, uint64_t end)
{
zfs_sb_t *zsb = ZTOZSB(zp);
dmu_tx_t *tx;
rl_t *rl;
int error;
sa_bulk_attr_t bulk[2];
int count = 0;
/*
* We will change zp_size, lock the whole file.
*/
rl = zfs_range_lock(zp, 0, UINT64_MAX, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (end >= zp->z_size) {
zfs_range_unlock(rl);
return (0);
}
error = dmu_free_long_range(zsb->z_os, zp->z_id, end, -1);
if (error) {
zfs_range_unlock(rl);
return (error);
}
top:
tx = dmu_tx_create(zsb->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
error = dmu_tx_assign(tx, TXG_NOWAIT);
if (error) {
if (error == ERESTART) {
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
zfs_range_unlock(rl);
return (error);
}
zp->z_size = end;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb),
NULL, &zp->z_size, sizeof (zp->z_size));
if (end == 0) {
zp->z_pflags &= ~ZFS_SPARSE;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb),
NULL, &zp->z_pflags, 8);
}
VERIFY(sa_bulk_update(zp->z_sa_hdl, bulk, count, tx) == 0);
dmu_tx_commit(tx);
zfs_range_unlock(rl);
return (0);
}
