/* ARGSUSED */
int
zfsctl_snapdir_mkdir(struct inode *dip, char *dirname, vattr_t *vap,
struct inode **ipp, cred_t *cr, int flags)
{
zfs_sb_t *zsb = ITOZSB(dip);
char *dsname;
int error;
dsname = kmem_alloc(MAXNAMELEN, KM_SLEEP);
if (snapshot_namecheck(dirname, NULL, NULL) != 0) {
error = EILSEQ;
goto out;
}
dmu_objset_name(zsb->z_os, dsname);
error = zfs_secpolicy_snapshot_perms(dsname, cr);
if (error)
goto out;
if (error == 0) {
error = dmu_objset_snapshot(dsname, dirname,
NULL, NULL, B_FALSE, B_FALSE, -1);
if (error)
goto out;
error = zfsctl_snapdir_lookup(dip, dirname, ipp,
0, cr, NULL, NULL);
}
out:
kmem_free(dsname, MAXNAMELEN);
return (error);
}
/* ARGSUSED */
static int
zfsctl_snapdir_mkdir(vnode_t *dvp, char *dirname, vattr_t *vap, vnode_t **vpp,
cred_t *cr, caller_context_t *cc, int flags, vsecattr_t *vsecp)
{
zfsvfs_t *zfsvfs = dvp->v_vfsp->vfs_data;
char name[MAXNAMELEN];
int err;
static enum symfollow follow = NO_FOLLOW;
static enum uio_seg seg = UIO_SYSSPACE;
if (snapshot_namecheck(dirname, NULL, NULL) != 0)
return (EILSEQ);
dmu_objset_name(zfsvfs->z_os, name);
*vpp = NULL;
err = zfs_secpolicy_snapshot_perms(name, cr);
if (err)
return (err);
if (err == 0) {
err = dmu_objset_snapshot(name, dirname, NULL, B_FALSE);
if (err)
return (err);
err = lookupnameat(dirname, seg, follow, NULL, vpp, dvp);
}
return (err);
}
void
zfs_fs_name_fn(void *ptr, char *osname)
{
zfsvfs_t *zfsvfs = NULL;
ASSERT(ptr);
zfsvfs = (zfsvfs_t *) ptr;
dmu_objset_name(zfsvfs->z_os, osname);
}
static int
zfsctl_snapshot_zname(vnode_t *vp, const char *name, int len, char *zname)
{
objset_t *os = ((zfsvfs_t *)((vp)->v_vfsp->vfs_data))->z_os;
if (snapshot_namecheck(name, NULL, NULL) != 0)
return (EILSEQ);
dmu_objset_name(os, zname);
if (strlen(zname) + 1 + strlen(name) >= len)
return (ENAMETOOLONG);
(void) strcat(zname, "@");
(void) strcat(zname, name);
return (0);
}
/*
* Block out VOPs and close zfsvfs_t::z_os
*
* Note, if successful, then we return with the 'z_teardown_lock' and
* 'z_teardown_inactive_lock' write held.
*/
int
zfs_suspend_fs(zfsvfs_t *zfsvfs, char *name, int *mode)
{
int error;
if ((error = zfsvfs_teardown(zfsvfs, B_FALSE)) != 0)
return (error);
*mode = zfsvfs->z_os->os_mode;
dmu_objset_name(zfsvfs->z_os, name);
dmu_objset_close(zfsvfs->z_os);
return (0);
}
static int
zfsctl_snapshot_zname(struct inode *ip, const char *name, int len, char *zname)
{
objset_t *os = ITOZSB(ip)->z_os;
if (snapshot_namecheck(name, NULL, NULL) != 0)
return (EILSEQ);
dmu_objset_name(os, zname);
if ((strlen(zname) + 1 + strlen(name)) >= len)
return (ENAMETOOLONG);
(void) strcat(zname, "@");
(void) strcat(zname, name);
return (0);
}
/*ARGSUSED*/
static int
zfsctl_snapdir_rename(vnode_t *sdvp, char *snm, vnode_t *tdvp, char *tnm,
cred_t *cr, caller_context_t *ct, int flags)
{
zfsctl_snapdir_t *sdp = sdvp->v_data;
zfs_snapentry_t search, *sep;
zfsvfs_t *zfsvfs;
avl_index_t where;
char from[ZFS_MAX_DATASET_NAME_LEN], to[ZFS_MAX_DATASET_NAME_LEN];
char real[ZFS_MAX_DATASET_NAME_LEN], fsname[ZFS_MAX_DATASET_NAME_LEN];
int err;
zfsvfs = sdvp->v_vfsp->vfs_data;
ZFS_ENTER(zfsvfs);
if ((flags & FIGNORECASE) || zfsvfs->z_case == ZFS_CASE_INSENSITIVE) {
err = dmu_snapshot_realname(zfsvfs->z_os, snm, real,
sizeof (real), NULL);
if (err == 0) {
snm = real;
} else if (err != ENOTSUP) {
ZFS_EXIT(zfsvfs);
return (err);
}
}
ZFS_EXIT(zfsvfs);
dmu_objset_name(zfsvfs->z_os, fsname);
err = zfsctl_snapshot_zname(sdvp, snm, sizeof (from), from);
if (err == 0)
err = zfsctl_snapshot_zname(tdvp, tnm, sizeof (to), to);
if (err == 0)
err = zfs_secpolicy_rename_perms(from, to, cr);
if (err != 0)
return (err);
/*
* Cannot move snapshots out of the snapdir.
*/
if (sdvp != tdvp)
return (SET_ERROR(EINVAL));
if (strcmp(snm, tnm) == 0)
return (0);
mutex_enter(&sdp->sd_lock);
search.se_name = (char *)snm;
if ((sep = avl_find(&sdp->sd_snaps, &search, &where)) == NULL) {
mutex_exit(&sdp->sd_lock);
return (SET_ERROR(ENOENT));
}
err = dsl_dataset_rename_snapshot(fsname, snm, tnm, B_FALSE);
if (err == 0)
zfsctl_rename_snap(sdp, sep, tnm);
mutex_exit(&sdp->sd_lock);
return (err);
}
static int
zfs_register_callbacks(vfs_t *vfsp)
{
struct dsl_dataset *ds = NULL;
objset_t *os = NULL;
zfsvfs_t *zfsvfs = NULL;
uint64_t nbmand;
int readonly, do_readonly = FALSE;
int setuid, do_setuid = FALSE;
int exec, do_exec = FALSE;
int xattr, do_xattr = FALSE;
int atime, do_atime = FALSE;
int error = 0;
ASSERT(vfsp);
zfsvfs = vfsp->vfs_data;
ASSERT(zfsvfs);
os = zfsvfs->z_os;
/*
* This function can be called for a snapshot when we update snapshot's
* mount point, which isn't really supported.
*/
if (dmu_objset_is_snapshot(os))
return (EOPNOTSUPP);
/*
* The act of registering our callbacks will destroy any mount
* options we may have. In order to enable temporary overrides
* of mount options, we stash away the current values and
* restore them after we register the callbacks.
*/
if (vfs_optionisset(vfsp, MNTOPT_RO, NULL)) {
readonly = B_TRUE;
do_readonly = B_TRUE;
} else if (vfs_optionisset(vfsp, MNTOPT_RW, NULL)) {
readonly = B_FALSE;
do_readonly = B_TRUE;
}
if (vfs_optionisset(vfsp, MNTOPT_NOSUID, NULL)) {
setuid = B_FALSE;
do_setuid = B_TRUE;
} else {
if (vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) {
setuid = B_FALSE;
do_setuid = B_TRUE;
} else if (vfs_optionisset(vfsp, MNTOPT_SETUID, NULL)) {
setuid = B_TRUE;
do_setuid = B_TRUE;
}
}
if (vfs_optionisset(vfsp, MNTOPT_NOEXEC, NULL)) {
exec = B_FALSE;
do_exec = B_TRUE;
} else if (vfs_optionisset(vfsp, MNTOPT_EXEC, NULL)) {
exec = B_TRUE;
do_exec = B_TRUE;
}
if (vfs_optionisset(vfsp, MNTOPT_NOXATTR, NULL)) {
xattr = B_FALSE;
do_xattr = B_TRUE;
} else if (vfs_optionisset(vfsp, MNTOPT_XATTR, NULL)) {
xattr = B_TRUE;
do_xattr = B_TRUE;
}
if (vfs_optionisset(vfsp, MNTOPT_NOATIME, NULL)) {
atime = B_FALSE;
do_atime = B_TRUE;
} else if (vfs_optionisset(vfsp, MNTOPT_ATIME, NULL)) {
atime = B_TRUE;
do_atime = B_TRUE;
}
/*
* nbmand is a special property. It can only be changed at
* mount time.
*
* This is weird, but it is documented to only be changeable
* at mount time.
*/
if (vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) {
nbmand = B_FALSE;
} else if (vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL)) {
nbmand = B_TRUE;
} else {
char osname[MAXNAMELEN];
dmu_objset_name(os, osname);
if (error = dsl_prop_get_integer(osname, "nbmand", &nbmand,
NULL)) {
return (error);
}
}
/*
* Register property callbacks.
*
* It would probably be fine to just check for i/o error from
* the first prop_register(), but I guess I like to go
* overboard...
*/
ds = dmu_objset_ds(os);
error = dsl_prop_register(ds, "atime", atime_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"xattr", xattr_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"recordsize", blksz_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"readonly", readonly_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"setuid", setuid_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"exec", exec_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"snapdir", snapdir_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"aclmode", acl_mode_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"aclinherit", acl_inherit_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,
"vscan", vscan_changed_cb, zfsvfs);
if (error)
goto unregister;
/*
* Invoke our callbacks to restore temporary mount options.
*/
if (do_readonly)
readonly_changed_cb(zfsvfs, readonly);
if (do_setuid)
setuid_changed_cb(zfsvfs, setuid);
if (do_exec)
exec_changed_cb(zfsvfs, exec);
if (do_xattr)
xattr_changed_cb(zfsvfs, xattr);
if (do_atime)
atime_changed_cb(zfsvfs, atime);
nbmand_changed_cb(zfsvfs, nbmand);
return (0);
unregister:
/*
* We may attempt to unregister some callbacks that are not
* registered, but this is OK; it will simply return ENOMSG,
* which we will ignore.
*/
(void) dsl_prop_unregister(ds, "atime", atime_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "xattr", xattr_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "recordsize", blksz_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "readonly", readonly_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "setuid", setuid_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "exec", exec_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "snapdir", snapdir_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "aclmode", acl_mode_changed_cb, zfsvfs);
(void) dsl_prop_unregister(ds, "aclinherit", acl_inherit_changed_cb,
zfsvfs);
(void) dsl_prop_unregister(ds, "vscan", vscan_changed_cb, zfsvfs);
return (error);
}
static void
zil_replay_log_record(zilog_t *zilog, lr_t *lr, void *zra, uint64_t claim_txg)
{
zil_replay_arg_t *zr = zra;
const zil_header_t *zh = zilog->zl_header;
uint64_t reclen = lr->lrc_reclen;
uint64_t txtype = lr->lrc_txtype;
char *name;
int pass, error;
if (!zilog->zl_replay) /* giving up */
return;
if (lr->lrc_txg < claim_txg) /* already committed */
return;
if (lr->lrc_seq <= zh->zh_replay_seq) /* already replayed */
return;
/* Strip case-insensitive bit, still present in log record */
txtype &= ~TX_CI;
if (txtype == 0 || txtype >= TX_MAX_TYPE) {
error = EINVAL;
goto bad;
}
/*
* Make a copy of the data so we can revise and extend it.
*/
bcopy(lr, zr->zr_lrbuf, reclen);
/*
* The log block containing this lr may have been byteswapped
* so that we can easily examine common fields like lrc_txtype.
* However, the log is a mix of different data types, and only the
* replay vectors know how to byteswap their records. Therefore, if
* the lr was byteswapped, undo it before invoking the replay vector.
*/
if (zr->zr_byteswap)
byteswap_uint64_array(zr->zr_lrbuf, reclen);
/*
* If this is a TX_WRITE with a blkptr, suck in the data.
*/
if (txtype == TX_WRITE && reclen == sizeof (lr_write_t)) {
lr_write_t *lrw = (lr_write_t *)lr;
blkptr_t *wbp = &lrw->lr_blkptr;
uint64_t wlen = lrw->lr_length;
char *wbuf = zr->zr_lrbuf + reclen;
if (BP_IS_HOLE(wbp)) { /* compressed to a hole */
bzero(wbuf, wlen);
} else {
/*
* A subsequent write may have overwritten this block,
* in which case wbp may have been been freed and
* reallocated, and our read of wbp may fail with a
* checksum error. We can safely ignore this because
* the later write will provide the correct data.
*/
zbookmark_t zb;
zb.zb_objset = dmu_objset_id(zilog->zl_os);
zb.zb_object = lrw->lr_foid;
zb.zb_level = -1;
zb.zb_blkid = lrw->lr_offset / BP_GET_LSIZE(wbp);
(void) zio_wait(zio_read(NULL, zilog->zl_spa,
wbp, wbuf, BP_GET_LSIZE(wbp), NULL, NULL,
ZIO_PRIORITY_SYNC_READ,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &zb));
(void) memmove(wbuf, wbuf + lrw->lr_blkoff, wlen);
}
}
/*
* We must now do two things atomically: replay this log record,
* and update the log header sequence number to reflect the fact that
* we did so. At the end of each replay function the sequence number
* is updated if we are in replay mode.
*/
for (pass = 1; pass <= 2; pass++) {
zilog->zl_replaying_seq = lr->lrc_seq;
/* Only byteswap (if needed) on the 1st pass. */
error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lrbuf,
zr->zr_byteswap && pass == 1);
if (!error)
return;
/*
* The DMU's dnode layer doesn't see removes until the txg
* commits, so a subsequent claim can spuriously fail with
* EEXIST. So if we receive any error we try syncing out
* any removes then retry the transaction.
*/
if (pass == 1)
txg_wait_synced(spa_get_dsl(zilog->zl_spa), 0);
}
bad:
ASSERT(error);
name = kmem_alloc(MAXNAMELEN, KM_SLEEP);
dmu_objset_name(zr->zr_os, name);
cmn_err(CE_WARN, "ZFS replay transaction error %d, "
"dataset %s, seq 0x%llx, txtype %llu %s\n",
error, name, (u_longlong_t)lr->lrc_seq, (u_longlong_t)txtype,
(lr->lrc_txtype & TX_CI) ? "CI" : "");
zilog->zl_replay = B_FALSE;
kmem_free(name, MAXNAMELEN);
}
static void
zil_replay_log_record(zilog_t *zilog, lr_t *lr, void *zra, uint64_t claim_txg)
{
zil_replay_arg_t *zr = zra;
const zil_header_t *zh = zilog->zl_header;
uint64_t reclen = lr->lrc_reclen;
uint64_t txtype = lr->lrc_txtype;
char *name;
int pass, error;
if (!zilog->zl_replay) /* giving up */
return;
if (lr->lrc_txg < claim_txg) /* already committed */
return;
if (lr->lrc_seq <= zh->zh_replay_seq) /* already replayed */
return;
/* Strip case-insensitive bit, still present in log record */
txtype &= ~TX_CI;
if (txtype == 0 || txtype >= TX_MAX_TYPE) {
error = EINVAL;
goto bad;
}
/*
* Make a copy of the data so we can revise and extend it.
*/
bcopy(lr, zr->zr_lrbuf, reclen);
/*
* The log block containing this lr may have been byteswapped
* so that we can easily examine common fields like lrc_txtype.
* However, the log is a mix of different data types, and only the
* replay vectors know how to byteswap their records. Therefore, if
* the lr was byteswapped, undo it before invoking the replay vector.
*/
if (zr->zr_byteswap)
byteswap_uint64_array(zr->zr_lrbuf, reclen);
/*
* We must now do two things atomically: replay this log record,
* and update the log header sequence number to reflect the fact that
* we did so. At the end of each replay function the sequence number
* is updated if we are in replay mode.
*/
for (pass = 1; pass <= 2; pass++) {
zilog->zl_replaying_seq = lr->lrc_seq;
/* Only byteswap (if needed) on the 1st pass. */
error = zr->zr_replay[txtype](zr->zr_arg, zr->zr_lrbuf,
zr->zr_byteswap && pass == 1);
if (!error)
return;
/*
* The DMU's dnode layer doesn't see removes until the txg
* commits, so a subsequent claim can spuriously fail with
* EEXIST. So if we receive any error we try syncing out
* any removes then retry the transaction.
*/
if (pass == 1)
txg_wait_synced(spa_get_dsl(zilog->zl_spa), 0);
}
bad:
ASSERT(error);
name = kmem_alloc(MAXNAMELEN, KM_SLEEP);
dmu_objset_name(zr->zr_os, name);
cmn_err(CE_WARN, "ZFS replay transaction error %d, "
"dataset %s, seq 0x%llx, txtype %llu %s\n",
error, name, (u_longlong_t)lr->lrc_seq, (u_longlong_t)txtype,
(lr->lrc_txtype & TX_CI) ? "CI" : "");
zilog->zl_replay = B_FALSE;
kmem_free(name, MAXNAMELEN);
}