static int
zfs_sa_setup(objset_t *osp, sa_attr_type_t **sa_table)
{
uint64_t sa_obj = 0;
int error;
error = zap_lookup(osp, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj);
if (error != 0 && error != ENOENT)
return (error);
error = sa_setup(osp, sa_obj, zfs_attr_table, ZPL_END, sa_table);
return (error);
}
int
zfs_sb_create(const char *osname, zfs_mntopts_t *zmo, zfs_sb_t **zsbp)
{
objset_t *os;
zfs_sb_t *zsb;
uint64_t zval;
int i, error;
uint64_t sa_obj;
zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_SLEEP);
/*
* We claim to always be readonly so we can open snapshots;
* other ZPL code will prevent us from writing to snapshots.
*/
error = dmu_objset_own(osname, DMU_OST_ZFS, B_TRUE, zsb, &os);
if (error) {
kmem_free(zsb, sizeof (zfs_sb_t));
return (error);
}
/*
* Optional temporary mount options, free'd in zfs_sb_free().
*/
zsb->z_mntopts = (zmo ? zmo : zfs_mntopts_alloc());
/*
* Initialize the zfs-specific filesystem structure.
* Should probably make this a kmem cache, shuffle fields,
* and just bzero up to z_hold_mtx[].
*/
zsb->z_sb = NULL;
zsb->z_parent = zsb;
zsb->z_max_blksz = SPA_OLD_MAXBLOCKSIZE;
zsb->z_show_ctldir = ZFS_SNAPDIR_VISIBLE;
zsb->z_os = os;
error = zfs_get_zplprop(os, ZFS_PROP_VERSION, &zsb->z_version);
if (error) {
goto out;
} else if (zsb->z_version > ZPL_VERSION) {
error = SET_ERROR(ENOTSUP);
goto out;
}
if ((error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &zval)) != 0)
goto out;
zsb->z_norm = (int)zval;
if ((error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &zval)) != 0)
goto out;
zsb->z_utf8 = (zval != 0);
if ((error = zfs_get_zplprop(os, ZFS_PROP_CASE, &zval)) != 0)
goto out;
zsb->z_case = (uint_t)zval;
if ((error = zfs_get_zplprop(os, ZFS_PROP_ACLTYPE, &zval)) != 0)
goto out;
zsb->z_acl_type = (uint_t)zval;
/*
* Fold case on file systems that are always or sometimes case
* insensitive.
*/
if (zsb->z_case == ZFS_CASE_INSENSITIVE ||
zsb->z_case == ZFS_CASE_MIXED)
zsb->z_norm |= U8_TEXTPREP_TOUPPER;
zsb->z_use_fuids = USE_FUIDS(zsb->z_version, zsb->z_os);
zsb->z_use_sa = USE_SA(zsb->z_version, zsb->z_os);
if (zsb->z_use_sa) {
/* should either have both of these objects or none */
error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1,
&sa_obj);
if (error)
goto out;
error = zfs_get_zplprop(os, ZFS_PROP_XATTR, &zval);
if ((error == 0) && (zval == ZFS_XATTR_SA))
zsb->z_xattr_sa = B_TRUE;
} else {
/*
* Pre SA versions file systems should never touch
* either the attribute registration or layout objects.
*/
sa_obj = 0;
}
error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END,
&zsb->z_attr_table);
if (error)
goto out;
if (zsb->z_version >= ZPL_VERSION_SA)
sa_register_update_callback(os, zfs_sa_upgrade);
error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1,
&zsb->z_root);
if (error)
goto out;
ASSERT(zsb->z_root != 0);
error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_UNLINKED_SET, 8, 1,
&zsb->z_unlinkedobj);
if (error)
goto out;
error = zap_lookup(os, MASTER_NODE_OBJ,
zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA],
8, 1, &zsb->z_userquota_obj);
if (error && error != ENOENT)
goto out;
error = zap_lookup(os, MASTER_NODE_OBJ,
zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA],
8, 1, &zsb->z_groupquota_obj);
if (error && error != ENOENT)
goto out;
error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, 8, 1,
&zsb->z_fuid_obj);
if (error && error != ENOENT)
goto out;
error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1,
&zsb->z_shares_dir);
if (error && error != ENOENT)
goto out;
mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL);
mutex_init(&zsb->z_lock, NULL, MUTEX_DEFAULT, NULL);
list_create(&zsb->z_all_znodes, sizeof (znode_t),
offsetof(znode_t, z_link_node));
rrm_init(&zsb->z_teardown_lock, B_FALSE);
rw_init(&zsb->z_teardown_inactive_lock, NULL, RW_DEFAULT, NULL);
rw_init(&zsb->z_fuid_lock, NULL, RW_DEFAULT, NULL);
zsb->z_hold_mtx = vmem_zalloc(sizeof (kmutex_t) * ZFS_OBJ_MTX_SZ,
KM_SLEEP);
for (i = 0; i != ZFS_OBJ_MTX_SZ; i++)
mutex_init(&zsb->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL);
*zsbp = zsb;
return (0);
out:
dmu_objset_disown(os, zsb);
*zsbp = NULL;
vmem_free(zsb->z_hold_mtx, sizeof (kmutex_t) * ZFS_OBJ_MTX_SZ);
kmem_free(zsb, sizeof (zfs_sb_t));
return (error);
}
/*
* Reopen zfs_sb_t and release VFS ops.
*/
int
zfs_resume_fs(zfs_sb_t *zsb, const char *osname)
{
int err, err2;
znode_t *zp;
uint64_t sa_obj = 0;
ASSERT(RRM_WRITE_HELD(&zsb->z_teardown_lock));
ASSERT(RW_WRITE_HELD(&zsb->z_teardown_inactive_lock));
/*
* We already own this, so just hold and rele it to update the
* objset_t, as the one we had before may have been evicted.
*/
VERIFY0(dmu_objset_hold(osname, zsb, &zsb->z_os));
VERIFY3P(zsb->z_os->os_dsl_dataset->ds_owner, ==, zsb);
VERIFY(dsl_dataset_long_held(zsb->z_os->os_dsl_dataset));
dmu_objset_rele(zsb->z_os, zsb);
/*
* Make sure version hasn't changed
*/
err = zfs_get_zplprop(zsb->z_os, ZFS_PROP_VERSION,
&zsb->z_version);
if (err)
goto bail;
err = zap_lookup(zsb->z_os, MASTER_NODE_OBJ,
ZFS_SA_ATTRS, 8, 1, &sa_obj);
if (err && zsb->z_version >= ZPL_VERSION_SA)
goto bail;
if ((err = sa_setup(zsb->z_os, sa_obj,
zfs_attr_table, ZPL_END, &zsb->z_attr_table)) != 0)
goto bail;
if (zsb->z_version >= ZPL_VERSION_SA)
sa_register_update_callback(zsb->z_os,
zfs_sa_upgrade);
VERIFY(zfs_sb_setup(zsb, B_FALSE) == 0);
zfs_set_fuid_feature(zsb);
zsb->z_rollback_time = jiffies;
/*
* Attempt to re-establish all the active inodes with their
* dbufs. If a zfs_rezget() fails, then we unhash the inode
* and mark it stale. This prevents a collision if a new
* inode/object is created which must use the same inode
* number. The stale inode will be be released when the
* VFS prunes the dentry holding the remaining references
* on the stale inode.
*/
mutex_enter(&zsb->z_znodes_lock);
for (zp = list_head(&zsb->z_all_znodes); zp;
zp = list_next(&zsb->z_all_znodes, zp)) {
err2 = zfs_rezget(zp);
if (err2) {
remove_inode_hash(ZTOI(zp));
zp->z_is_stale = B_TRUE;
}
}
mutex_exit(&zsb->z_znodes_lock);
bail:
/* release the VFS ops */
rw_exit(&zsb->z_teardown_inactive_lock);
rrm_exit(&zsb->z_teardown_lock, FTAG);
if (err) {
/*
* Since we couldn't setup the sa framework, try to force
* unmount this file system.
*/
if (zsb->z_os)
(void) zfs_umount(zsb->z_sb);
}
return (err);
}
/*
* Reopen zfs_sb_t and release VFS ops.
*/
int
zfs_resume_fs(zfs_sb_t *zsb, const char *osname)
{
int err, err2;
ASSERT(RRW_WRITE_HELD(&zsb->z_teardown_lock));
ASSERT(RW_WRITE_HELD(&zsb->z_teardown_inactive_lock));
err = dmu_objset_own(osname, DMU_OST_ZFS, B_FALSE, zsb, &zsb->z_os);
if (err) {
zsb->z_os = NULL;
} else {
znode_t *zp;
uint64_t sa_obj = 0;
err2 = zap_lookup(zsb->z_os, MASTER_NODE_OBJ,
ZFS_SA_ATTRS, 8, 1, &sa_obj);
if ((err || err2) && zsb->z_version >= ZPL_VERSION_SA)
goto bail;
if ((err = sa_setup(zsb->z_os, sa_obj,
zfs_attr_table, ZPL_END, &zsb->z_attr_table)) != 0)
goto bail;
VERIFY(zfs_sb_setup(zsb, B_FALSE) == 0);
zsb->z_rollback_time = jiffies;
/*
* Attempt to re-establish all the active inodes with their
* dbufs. If a zfs_rezget() fails, then we unhash the inode
* and mark it stale. This prevents a collision if a new
* inode/object is created which must use the same inode
* number. The stale inode will be be released when the
* VFS prunes the dentry holding the remaining references
* on the stale inode.
*/
mutex_enter(&zsb->z_znodes_lock);
for (zp = list_head(&zsb->z_all_znodes); zp;
zp = list_next(&zsb->z_all_znodes, zp)) {
err2 = zfs_rezget(zp);
if (err2) {
remove_inode_hash(ZTOI(zp));
zp->z_is_stale = B_TRUE;
}
}
mutex_exit(&zsb->z_znodes_lock);
}
bail:
/* release the VFS ops */
rw_exit(&zsb->z_teardown_inactive_lock);
rrw_exit(&zsb->z_teardown_lock, FTAG);
if (err) {
/*
* Since we couldn't reopen zfs_sb_t or, setup the
* sa framework, force unmount this file system.
*/
if (zsb->z_os)
(void) zfs_umount(zsb->z_sb);
}
return (err);
}
void
zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx)
{
struct super_block *sb;
zfs_sb_t *zsb;
uint64_t moid, obj, sa_obj, version;
uint64_t sense = ZFS_CASE_SENSITIVE;
uint64_t norm = 0;
nvpair_t *elem;
int error;
int i;
znode_t *rootzp = NULL;
vattr_t vattr;
znode_t *zp;
zfs_acl_ids_t acl_ids;
/*
* First attempt to create master node.
*/
/*
* In an empty objset, there are no blocks to read and thus
* there can be no i/o errors (which we assert below).
*/
moid = MASTER_NODE_OBJ;
error = zap_create_claim(os, moid, DMU_OT_MASTER_NODE,
DMU_OT_NONE, 0, tx);
ASSERT(error == 0);
/*
* Set starting attributes.
*/
version = zfs_zpl_version_map(spa_version(dmu_objset_spa(os)));
elem = NULL;
while ((elem = nvlist_next_nvpair(zplprops, elem)) != NULL) {
/* For the moment we expect all zpl props to be uint64_ts */
uint64_t val;
char *name;
ASSERT(nvpair_type(elem) == DATA_TYPE_UINT64);
VERIFY(nvpair_value_uint64(elem, &val) == 0);
name = nvpair_name(elem);
if (strcmp(name, zfs_prop_to_name(ZFS_PROP_VERSION)) == 0) {
if (val < version)
version = val;
} else {
error = zap_update(os, moid, name, 8, 1, &val, tx);
}
ASSERT(error == 0);
if (strcmp(name, zfs_prop_to_name(ZFS_PROP_NORMALIZE)) == 0)
norm = val;
else if (strcmp(name, zfs_prop_to_name(ZFS_PROP_CASE)) == 0)
sense = val;
}
ASSERT(version != 0);
error = zap_update(os, moid, ZPL_VERSION_STR, 8, 1, &version, tx);
/*
* Create zap object used for SA attribute registration
*/
if (version >= ZPL_VERSION_SA) {
sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE,
DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx);
ASSERT(error == 0);
} else {
sa_obj = 0;
}
/*
* Create a delete queue.
*/
obj = zap_create(os, DMU_OT_UNLINKED_SET, DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_UNLINKED_SET, 8, 1, &obj, tx);
ASSERT(error == 0);
/*
* Create root znode. Create minimal znode/inode/zsb/sb
* to allow zfs_mknode to work.
*/
vattr.va_mask = ATTR_MODE|ATTR_UID|ATTR_GID;
vattr.va_mode = S_IFDIR|0755;
vattr.va_uid = crgetuid(cr);
vattr.va_gid = crgetgid(cr);
rootzp = kmem_cache_alloc(znode_cache, KM_PUSHPAGE);
rootzp->z_moved = 0;
rootzp->z_unlinked = 0;
rootzp->z_atime_dirty = 0;
rootzp->z_is_sa = USE_SA(version, os);
zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_PUSHPAGE | KM_NODEBUG);
zsb->z_os = os;
zsb->z_parent = zsb;
zsb->z_version = version;
zsb->z_use_fuids = USE_FUIDS(version, os);
zsb->z_use_sa = USE_SA(version, os);
zsb->z_norm = norm;
sb = kmem_zalloc(sizeof (struct super_block), KM_PUSHPAGE);
sb->s_fs_info = zsb;
ZTOI(rootzp)->i_sb = sb;
error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END,
&zsb->z_attr_table);
ASSERT(error == 0);
/*
* Fold case on file systems that are always or sometimes case
* insensitive.
*/
if (sense == ZFS_CASE_INSENSITIVE || sense == ZFS_CASE_MIXED)
zsb->z_norm |= U8_TEXTPREP_TOUPPER;
mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL);
list_create(&zsb->z_all_znodes, sizeof (znode_t),
offsetof(znode_t, z_link_node));
for (i = 0; i != ZFS_OBJ_MTX_SZ; i++)
mutex_init(&zsb->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL);
VERIFY(0 == zfs_acl_ids_create(rootzp, IS_ROOT_NODE, &vattr,
cr, NULL, &acl_ids));
zfs_mknode(rootzp, &vattr, tx, cr, IS_ROOT_NODE, &zp, &acl_ids);
ASSERT3P(zp, ==, rootzp);
error = zap_add(os, moid, ZFS_ROOT_OBJ, 8, 1, &rootzp->z_id, tx);
ASSERT(error == 0);
zfs_acl_ids_free(&acl_ids);
atomic_set(&ZTOI(rootzp)->i_count, 0);
sa_handle_destroy(rootzp->z_sa_hdl);
kmem_cache_free(znode_cache, rootzp);
/*
* Create shares directory
*/
error = zfs_create_share_dir(zsb, tx);
ASSERT(error == 0);
for (i = 0; i != ZFS_OBJ_MTX_SZ; i++)
mutex_destroy(&zsb->z_hold_mtx[i]);
kmem_free(sb, sizeof (struct super_block));
kmem_free(zsb, sizeof (zfs_sb_t));
}