ssize_t
zpl_xattr_list(struct dentry *dentry, char *buffer, size_t buffer_size)
{
znode_t *zp = ITOZ(dentry->d_inode);
zfs_sb_t *zsb = ZTOZSB(zp);
xattr_filldir_t xf = { buffer_size, 0, buffer, dentry->d_inode };
cred_t *cr = CRED();
fstrans_cookie_t cookie;
int error = 0;
crhold(cr);
cookie = spl_fstrans_mark();
rrm_enter_read(&(zsb)->z_teardown_lock, FTAG);
rw_enter(&zp->z_xattr_lock, RW_READER);
if (zsb->z_use_sa && zp->z_is_sa) {
error = zpl_xattr_list_sa(&xf);
if (error)
goto out;
}
error = zpl_xattr_list_dir(&xf, cr);
if (error)
goto out;
error = xf.offset;
out:
rw_exit(&zp->z_xattr_lock);
rrm_exit(&(zsb)->z_teardown_lock, FTAG);
spl_fstrans_unmark(cookie);
crfree(cr);
return (error);
}
static int
zpl_xattr_set(struct inode *ip, const char *name, const void *value,
size_t size, int flags)
{
znode_t *zp = ITOZ(ip);
zfs_sb_t *zsb = ZTOZSB(zp);
cred_t *cr = CRED();
fstrans_cookie_t cookie;
int error;
crhold(cr);
cookie = spl_fstrans_mark();
rrm_enter_read(&(zsb)->z_teardown_lock, FTAG);
rw_enter(&ITOZ(ip)->z_xattr_lock, RW_WRITER);
/*
* Before setting the xattr check to see if it already exists.
* This is done to ensure the following optional flags are honored.
*
* XATTR_CREATE: fail if xattr already exists
* XATTR_REPLACE: fail if xattr does not exist
*/
error = __zpl_xattr_get(ip, name, NULL, 0, cr);
if (error < 0) {
if (error != -ENODATA)
goto out;
if (flags & XATTR_REPLACE)
goto out;
/* The xattr to be removed already doesn't exist */
error = 0;
if (value == NULL)
goto out;
} else {
error = -EEXIST;
if (flags & XATTR_CREATE)
goto out;
}
/* Preferentially store the xattr as a SA for better performance */
if (zsb->z_use_sa && zsb->z_xattr_sa && zp->z_is_sa) {
error = zpl_xattr_set_sa(ip, name, value, size, flags, cr);
if (error == 0)
goto out;
}
error = zpl_xattr_set_dir(ip, name, value, size, flags, cr);
out:
rw_exit(&ITOZ(ip)->z_xattr_lock);
rrm_exit(&(zsb)->z_teardown_lock, FTAG);
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
zpl_xattr_get(struct inode *ip, const char *name, void *value, size_t size)
{
znode_t *zp = ITOZ(ip);
zfs_sb_t *zsb = ZTOZSB(zp);
cred_t *cr = CRED();
fstrans_cookie_t cookie;
int error;
crhold(cr);
cookie = spl_fstrans_mark();
rrm_enter_read(&(zsb)->z_teardown_lock, FTAG);
rw_enter(&zp->z_xattr_lock, RW_READER);
error = __zpl_xattr_get(ip, name, value, size, cr);
rw_exit(&zp->z_xattr_lock);
rrm_exit(&(zsb)->z_teardown_lock, FTAG);
spl_fstrans_unmark(cookie);
crfree(cr);
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);
}
/*
* Teardown the zfs_sb_t.
*
* Note, if 'unmounting' if FALSE, we return with the 'z_teardown_lock'
* and 'z_teardown_inactive_lock' held.
*/
int
zfs_sb_teardown(zfs_sb_t *zsb, boolean_t unmounting)
{
znode_t *zp;
/*
* If someone has not already unmounted this file system,
* drain the iput_taskq to ensure all active references to the
* zfs_sb_t have been handled only then can it be safely destroyed.
*/
if (zsb->z_os) {
/*
* If we're unmounting we have to wait for the list to
* drain completely.
*
* If we're not unmounting there's no guarantee the list
* will drain completely, but iputs run from the taskq
* may add the parents of dir-based xattrs to the taskq
* so we want to wait for these.
*
* We can safely read z_nr_znodes without locking because the
* VFS has already blocked operations which add to the
* z_all_znodes list and thus increment z_nr_znodes.
*/
int round = 0;
while (zsb->z_nr_znodes > 0) {
taskq_wait_outstanding(dsl_pool_iput_taskq(
dmu_objset_pool(zsb->z_os)), 0);
if (++round > 1 && !unmounting)
break;
}
}
rrm_enter(&zsb->z_teardown_lock, RW_WRITER, FTAG);
if (!unmounting) {
/*
* We purge the parent filesystem's super block as the
* parent filesystem and all of its snapshots have their
* inode's super block set to the parent's filesystem's
* super block. Note, 'z_parent' is self referential
* for non-snapshots.
*/
shrink_dcache_sb(zsb->z_parent->z_sb);
}
/*
* Close the zil. NB: Can't close the zil while zfs_inactive
* threads are blocked as zil_close can call zfs_inactive.
*/
if (zsb->z_log) {
zil_close(zsb->z_log);
zsb->z_log = NULL;
}
rw_enter(&zsb->z_teardown_inactive_lock, RW_WRITER);
/*
* If we are not unmounting (ie: online recv) and someone already
* unmounted this file system while we were doing the switcheroo,
* or a reopen of z_os failed then just bail out now.
*/
if (!unmounting && (zsb->z_unmounted || zsb->z_os == NULL)) {
rw_exit(&zsb->z_teardown_inactive_lock);
rrm_exit(&zsb->z_teardown_lock, FTAG);
return (SET_ERROR(EIO));
}
/*
* At this point there are no VFS ops active, and any new VFS ops
* will fail with EIO since we have z_teardown_lock for writer (only
* relevant for forced unmount).
*
* Release all holds on dbufs.
*/
if (!unmounting) {
mutex_enter(&zsb->z_znodes_lock);
for (zp = list_head(&zsb->z_all_znodes); zp != NULL;
zp = list_next(&zsb->z_all_znodes, zp)) {
if (zp->z_sa_hdl)
zfs_znode_dmu_fini(zp);
}
mutex_exit(&zsb->z_znodes_lock);
}
/*
* If we are unmounting, set the unmounted flag and let new VFS ops
* unblock. zfs_inactive will have the unmounted behavior, and all
* other VFS ops will fail with EIO.
*/
if (unmounting) {
zsb->z_unmounted = B_TRUE;
rrm_exit(&zsb->z_teardown_lock, FTAG);
rw_exit(&zsb->z_teardown_inactive_lock);
}
/*
* z_os will be NULL if there was an error in attempting to reopen
* zsb, so just return as the properties had already been
*
* unregistered and cached data had been evicted before.
*/
if (zsb->z_os == NULL)
return (0);
/*
* Unregister properties.
*/
zfs_unregister_callbacks(zsb);
/*
* Evict cached data
*/
if (dsl_dataset_is_dirty(dmu_objset_ds(zsb->z_os)) &&
!zfs_is_readonly(zsb))
txg_wait_synced(dmu_objset_pool(zsb->z_os), 0);
dmu_objset_evict_dbufs(zsb->z_os);
return (0);
}
static int
zpl_xattr_set(struct inode *ip, const char *name, const void *value,
size_t size, int flags)
{
znode_t *zp = ITOZ(ip);
zfs_sb_t *zsb = ZTOZSB(zp);
cred_t *cr = CRED();
fstrans_cookie_t cookie;
int where;
int error;
crhold(cr);
cookie = spl_fstrans_mark();
rrm_enter_read(&(zsb)->z_teardown_lock, FTAG);
rw_enter(&ITOZ(ip)->z_xattr_lock, RW_WRITER);
/*
* Before setting the xattr check to see if it already exists.
* This is done to ensure the following optional flags are honored.
*
* XATTR_CREATE: fail if xattr already exists
* XATTR_REPLACE: fail if xattr does not exist
*
* We also want to know if it resides in sa or dir, so we can make
* sure we don't end up with duplicate in both places.
*/
error = __zpl_xattr_where(ip, name, &where, cr);
if (error < 0) {
if (error != -ENODATA)
goto out;
if (flags & XATTR_REPLACE)
goto out;
/* The xattr to be removed already doesn't exist */
error = 0;
if (value == NULL)
goto out;
} else {
error = -EEXIST;
if (flags & XATTR_CREATE)
goto out;
}
/* Preferentially store the xattr as a SA for better performance */
if (zsb->z_use_sa && zp->z_is_sa &&
(zsb->z_xattr_sa || (value == NULL && where & XATTR_IN_SA))) {
error = zpl_xattr_set_sa(ip, name, value, size, flags, cr);
if (error == 0) {
/*
* Successfully put into SA, we need to clear the one
* in dir.
*/
if (where & XATTR_IN_DIR)
zpl_xattr_set_dir(ip, name, NULL, 0, 0, cr);
goto out;
}
}
error = zpl_xattr_set_dir(ip, name, value, size, flags, cr);
/*
* Successfully put into dir, we need to clear the one in SA.
*/
if (error == 0 && (where & XATTR_IN_SA))
zpl_xattr_set_sa(ip, name, NULL, 0, 0, cr);
out:
rw_exit(&ITOZ(ip)->z_xattr_lock);
rrm_exit(&(zsb)->z_teardown_lock, FTAG);
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}