/*ARGSUSED*/
int
zfs_umount(struct super_block *sb)
{
zfs_sb_t *zsb = sb->s_fs_info;
objset_t *os;
arc_remove_prune_callback(zsb->z_arc_prune);
VERIFY(zfs_sb_teardown(zsb, B_TRUE) == 0);
os = zsb->z_os;
bdi_destroy(sb->s_bdi);
/*
* z_os will be NULL if there was an error in
* attempting to reopen zsb.
*/
if (os != NULL) {
/*
* Unset the objset user_ptr.
*/
mutex_enter(&os->os_user_ptr_lock);
dmu_objset_set_user(os, NULL);
mutex_exit(&os->os_user_ptr_lock);
/*
* Finally release the objset
*/
dmu_objset_disown(os, zsb);
}
zfs_sb_free(zsb);
return (0);
}
static int
zfsvfs_setup(zfsvfs_t *zfsvfs, boolean_t mounting)
{
int error;
error = zfs_register_callbacks(zfsvfs->z_vfs);
if (error)
return (error);
/*
* Set the objset user_ptr to track its zfsvfs.
*/
mutex_enter(&zfsvfs->z_os->os->os_user_ptr_lock);
dmu_objset_set_user(zfsvfs->z_os, zfsvfs);
mutex_exit(&zfsvfs->z_os->os->os_user_ptr_lock);
/*
* If we are not mounting (ie: online recv), then we don't
* have to worry about replaying the log as we blocked all
* operations out since we closed the ZIL.
*/
if (mounting) {
boolean_t readonly;
/*
* During replay we remove the read only flag to
* allow replays to succeed.
*/
readonly = zfsvfs->z_vfs->vfs_flag & VFS_RDONLY;
zfsvfs->z_vfs->vfs_flag &= ~VFS_RDONLY;
/*
* Parse and replay the intent log.
*/
zil_replay(zfsvfs->z_os, zfsvfs, &zfsvfs->z_assign,
zfs_replay_vector, zfs_unlinked_drain);
zfs_unlinked_drain(zfsvfs);
zfsvfs->z_vfs->vfs_flag |= readonly; /* restore readonly bit */
}
if (!zil_disable)
zfsvfs->z_log = zil_open(zfsvfs->z_os, zfs_get_data);
return (0);
}
int
zfs_sb_setup(zfs_sb_t *zsb, boolean_t mounting)
{
int error;
error = zfs_register_callbacks(zsb);
if (error)
return (error);
/*
* Set the objset user_ptr to track its zsb.
*/
mutex_enter(&zsb->z_os->os_user_ptr_lock);
dmu_objset_set_user(zsb->z_os, zsb);
mutex_exit(&zsb->z_os->os_user_ptr_lock);
zsb->z_log = zil_open(zsb->z_os, zfs_get_data);
/*
* If we are not mounting (ie: online recv), then we don't
* have to worry about replaying the log as we blocked all
* operations out since we closed the ZIL.
*/
if (mounting) {
boolean_t readonly;
/*
* During replay we remove the read only flag to
* allow replays to succeed.
*/
readonly = zfs_is_readonly(zsb);
if (readonly != 0)
readonly_changed_cb(zsb, B_FALSE);
else
zfs_unlinked_drain(zsb);
/*
* Parse and replay the intent log.
*
* Because of ziltest, this must be done after
* zfs_unlinked_drain(). (Further note: ziltest
* doesn't use readonly mounts, where
* zfs_unlinked_drain() isn't called.) This is because
* ziltest causes spa_sync() to think it's committed,
* but actually it is not, so the intent log contains
* many txg's worth of changes.
*
* In particular, if object N is in the unlinked set in
* the last txg to actually sync, then it could be
* actually freed in a later txg and then reallocated
* in a yet later txg. This would write a "create
* object N" record to the intent log. Normally, this
* would be fine because the spa_sync() would have
* written out the fact that object N is free, before
* we could write the "create object N" intent log
* record.
*
* But when we are in ziltest mode, we advance the "open
* txg" without actually spa_sync()-ing the changes to
* disk. So we would see that object N is still
* allocated and in the unlinked set, and there is an
* intent log record saying to allocate it.
*/
if (spa_writeable(dmu_objset_spa(zsb->z_os))) {
if (zil_replay_disable) {
zil_destroy(zsb->z_log, B_FALSE);
} else {
zsb->z_replay = B_TRUE;
zil_replay(zsb->z_os, zsb,
zfs_replay_vector);
zsb->z_replay = B_FALSE;
}
}
/* restore readonly bit */
if (readonly != 0)
readonly_changed_cb(zsb, B_TRUE);
}
return (0);
}
int
zfs_domount(struct super_block *sb, zfs_mntopts_t *zmo, int silent)
{
const char *osname = zmo->z_osname;
zfs_sb_t *zsb;
struct inode *root_inode;
uint64_t recordsize;
int error;
error = zfs_sb_create(osname, zmo, &zsb);
if (error)
return (error);
if ((error = dsl_prop_get_integer(osname, "recordsize",
&recordsize, NULL)))
goto out;
zsb->z_sb = sb;
sb->s_fs_info = zsb;
sb->s_magic = ZFS_SUPER_MAGIC;
sb->s_maxbytes = MAX_LFS_FILESIZE;
sb->s_time_gran = 1;
sb->s_blocksize = recordsize;
sb->s_blocksize_bits = ilog2(recordsize);
zsb->z_bdi.ra_pages = 0;
sb->s_bdi = &zsb->z_bdi;
error = -zpl_bdi_setup_and_register(&zsb->z_bdi, "zfs");
if (error)
goto out;
/* Set callback operations for the file system. */
sb->s_op = &zpl_super_operations;
sb->s_xattr = zpl_xattr_handlers;
sb->s_export_op = &zpl_export_operations;
#ifdef HAVE_S_D_OP
sb->s_d_op = &zpl_dentry_operations;
#endif /* HAVE_S_D_OP */
/* Set features for file system. */
zfs_set_fuid_feature(zsb);
if (dmu_objset_is_snapshot(zsb->z_os)) {
uint64_t pval;
atime_changed_cb(zsb, B_FALSE);
readonly_changed_cb(zsb, B_TRUE);
if ((error = dsl_prop_get_integer(osname,
"xattr", &pval, NULL)))
goto out;
xattr_changed_cb(zsb, pval);
if ((error = dsl_prop_get_integer(osname,
"acltype", &pval, NULL)))
goto out;
acltype_changed_cb(zsb, pval);
zsb->z_issnap = B_TRUE;
zsb->z_os->os_sync = ZFS_SYNC_DISABLED;
zsb->z_snap_defer_time = jiffies;
mutex_enter(&zsb->z_os->os_user_ptr_lock);
dmu_objset_set_user(zsb->z_os, zsb);
mutex_exit(&zsb->z_os->os_user_ptr_lock);
} else {
error = zfs_sb_setup(zsb, B_TRUE);
}
/* Allocate a root inode for the filesystem. */
error = zfs_root(zsb, &root_inode);
if (error) {
(void) zfs_umount(sb);
goto out;
}
/* Allocate a root dentry for the filesystem */
sb->s_root = d_make_root(root_inode);
if (sb->s_root == NULL) {
(void) zfs_umount(sb);
error = SET_ERROR(ENOMEM);
goto out;
}
if (!zsb->z_issnap)
zfsctl_create(zsb);
zsb->z_arc_prune = arc_add_prune_callback(zpl_prune_sb, sb);
out:
if (error) {
dmu_objset_disown(zsb->z_os, zsb);
zfs_sb_free(zsb);
}
return (error);
}
int
zfs_domount(struct super_block *sb, void *data, int silent)
{
zpl_mount_data_t *zmd = data;
const char *osname = zmd->z_osname;
zfs_sb_t *zsb;
struct inode *root_inode;
uint64_t recordsize;
int error;
error = zfs_sb_create(osname, &zsb);
if (error)
return (error);
if ((error = dsl_prop_get_integer(osname, "recordsize",
&recordsize, NULL)))
goto out;
zsb->z_sb = sb;
sb->s_fs_info = zsb;
sb->s_magic = ZFS_SUPER_MAGIC;
sb->s_maxbytes = MAX_LFS_FILESIZE;
sb->s_time_gran = 1;
sb->s_blocksize = recordsize;
sb->s_blocksize_bits = ilog2(recordsize);
#ifdef HAVE_BDI
/*
* 2.6.32 API change,
* Added backing_device_info (BDI) per super block interfaces. A BDI
* must be configured when using a non-device backed filesystem for
* proper writeback. This is not required for older pdflush kernels.
*
* NOTE: Linux read-ahead is disabled in favor of zfs read-ahead.
*/
zsb->z_bdi.ra_pages = 0;
sb->s_bdi = &zsb->z_bdi;
error = -bdi_setup_and_register(&zsb->z_bdi, "zfs", BDI_CAP_MAP_COPY);
if (error)
goto out;
#endif /* HAVE_BDI */
/* Set callback operations for the file system. */
sb->s_op = &zpl_super_operations;
sb->s_xattr = zpl_xattr_handlers;
sb->s_export_op = &zpl_export_operations;
#ifdef HAVE_S_D_OP
sb->s_d_op = &zpl_dentry_operations;
#endif /* HAVE_S_D_OP */
/* Set features for file system. */
zfs_set_fuid_feature(zsb);
if (dmu_objset_is_snapshot(zsb->z_os)) {
uint64_t pval;
atime_changed_cb(zsb, B_FALSE);
readonly_changed_cb(zsb, B_TRUE);
if ((error = dsl_prop_get_integer(osname,"xattr",&pval,NULL)))
goto out;
xattr_changed_cb(zsb, pval);
zsb->z_issnap = B_TRUE;
zsb->z_os->os_sync = ZFS_SYNC_DISABLED;
mutex_enter(&zsb->z_os->os_user_ptr_lock);
dmu_objset_set_user(zsb->z_os, zsb);
mutex_exit(&zsb->z_os->os_user_ptr_lock);
} else {
error = zfs_sb_setup(zsb, B_TRUE);
}
/* Allocate a root inode for the filesystem. */
error = zfs_root(zsb, &root_inode);
if (error) {
(void) zfs_umount(sb);
goto out;
}
/* Allocate a root dentry for the filesystem */
sb->s_root = d_make_root(root_inode);
if (sb->s_root == NULL) {
(void) zfs_umount(sb);
error = ENOMEM;
goto out;
}
if (!zsb->z_issnap)
zfsctl_create(zsb);
out:
if (error) {
dmu_objset_disown(zsb->z_os, zsb);
zfs_sb_free(zsb);
}
return (error);
}
/*ARGSUSED*/
static int
zfs_umount(vfs_t *vfsp, int fflag)
{
zfsvfs_t *zfsvfs = vfsp->vfs_data;
objset_t *os;
cred_t *cr = curthread->td_ucred;
int ret;
ret = secpolicy_fs_unmount(cr, vfsp);
if (ret) {
ret = dsl_deleg_access((char *)refstr_value(vfsp->vfs_resource),
ZFS_DELEG_PERM_MOUNT, cr);
if (ret)
return (ret);
}
/*
* We purge the parent filesystem's vfsp as the parent filesystem
* and all of its snapshots have their vnode's v_vfsp set to the
* parent's filesystem's vfsp. Note, 'z_parent' is self
* referential for non-snapshots.
*/
(void) dnlc_purge_vfsp(zfsvfs->z_parent->z_vfs, 0);
/*
* Unmount any snapshots mounted under .zfs before unmounting the
* dataset itself.
*/
if (zfsvfs->z_ctldir != NULL) {
if ((ret = zfsctl_umount_snapshots(vfsp, fflag, cr)) != 0)
return (ret);
ret = vflush(vfsp, 0, 0, curthread);
ASSERT(ret == EBUSY);
if (!(fflag & MS_FORCE)) {
if (zfsvfs->z_ctldir->v_count > 1)
return (EBUSY);
ASSERT(zfsvfs->z_ctldir->v_count == 1);
}
zfsctl_destroy(zfsvfs);
ASSERT(zfsvfs->z_ctldir == NULL);
}
if (fflag & MS_FORCE) {
/*
* Mark file system as unmounted before calling
* vflush(FORCECLOSE). This way we ensure no future vnops
* will be called and risk operating on DOOMED vnodes.
*/
rrw_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG);
zfsvfs->z_unmounted = B_TRUE;
rrw_exit(&zfsvfs->z_teardown_lock, FTAG);
}
/*
* Flush all the files.
*/
ret = vflush(vfsp, 1, (fflag & MS_FORCE) ? FORCECLOSE : 0, curthread);
if (ret != 0) {
if (!zfsvfs->z_issnap) {
zfsctl_create(zfsvfs);
ASSERT(zfsvfs->z_ctldir != NULL);
}
return (ret);
}
if (!(fflag & MS_FORCE)) {
/*
* Check the number of active vnodes in the file system.
* Our count is maintained in the vfs structure, but the
* number is off by 1 to indicate a hold on the vfs
* structure itself.
*
* The '.zfs' directory maintains a reference of its
* own, and any active references underneath are
* reflected in the vnode count.
*/
if (zfsvfs->z_ctldir == NULL) {
if (vfsp->vfs_count > 1)
return (EBUSY);
} else {
if (vfsp->vfs_count > 2 ||
zfsvfs->z_ctldir->v_count > 1)
return (EBUSY);
}
} else {
MNT_ILOCK(vfsp);
vfsp->mnt_kern_flag |= MNTK_UNMOUNTF;
MNT_IUNLOCK(vfsp);
}
VERIFY(zfsvfs_teardown(zfsvfs, B_TRUE) == 0);
os = zfsvfs->z_os;
/*
* z_os will be NULL if there was an error in
* attempting to reopen zfsvfs.
*/
if (os != NULL) {
/*
* Unset the objset user_ptr.
*/
mutex_enter(&os->os->os_user_ptr_lock);
dmu_objset_set_user(os, NULL);
mutex_exit(&os->os->os_user_ptr_lock);
/*
* Finally release the objset
*/
dmu_objset_close(os);
}
/*
* We can now safely destroy the '.zfs' directory node.
*/
if (zfsvfs->z_ctldir != NULL)
zfsctl_destroy(zfsvfs);
if (zfsvfs->z_issnap) {
vnode_t *svp = vfsp->mnt_vnodecovered;
if (svp->v_count >= 2)
VN_RELE(svp);
}
zfs_freevfs(vfsp);
return (0);
}
