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));
}
/*
* Link zp into dl. Can only fail if zp has been unlinked.
*/
int
zfs_link_create(zfs_dirlock_t *dl, znode_t *zp, dmu_tx_t *tx, int flag)
{
znode_t *dzp = dl->dl_dzp;
zfs_sb_t *zsb = ZTOZSB(zp);
uint64_t value;
int zp_is_dir = S_ISDIR(ZTOI(zp)->i_mode);
sa_bulk_attr_t bulk[5];
uint64_t mtime[2], ctime[2];
int count = 0;
int error;
mutex_enter(&zp->z_lock);
if (!(flag & ZRENAMING)) {
if (zp->z_unlinked) { /* no new links to unlinked zp */
ASSERT(!(flag & (ZNEW | ZEXISTS)));
mutex_exit(&zp->z_lock);
return (ENOENT);
}
zp->z_links++;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL,
&zp->z_links, sizeof (zp->z_links));
}
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zsb), NULL,
&dzp->z_id, sizeof (dzp->z_id));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL,
&zp->z_pflags, sizeof (zp->z_pflags));
if (!(flag & ZNEW)) {
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL,
ctime, sizeof (ctime));
zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime,
ctime, B_TRUE);
}
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
mutex_exit(&zp->z_lock);
mutex_enter(&dzp->z_lock);
dzp->z_size++;
dzp->z_links += zp_is_dir;
count = 0;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL,
&dzp->z_size, sizeof (dzp->z_size));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL,
&dzp->z_links, sizeof (dzp->z_links));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL,
mtime, sizeof (mtime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL,
ctime, sizeof (ctime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL,
&dzp->z_pflags, sizeof (dzp->z_pflags));
zfs_tstamp_update_setup(dzp, CONTENT_MODIFIED, mtime, ctime, B_TRUE);
error = sa_bulk_update(dzp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
mutex_exit(&dzp->z_lock);
value = zfs_dirent(zp, zp->z_mode);
error = zap_add(ZTOZSB(zp)->z_os, dzp->z_id, dl->dl_name,
8, 1, &value, tx);
ASSERT(error == 0);
return (0);
}
void
zfs_sa_upgrade(sa_handle_t *hdl, dmu_tx_t *tx)
{
dmu_buf_t *db = sa_get_db(hdl);
znode_t *zp = sa_get_userdata(hdl);
zfs_sb_t *zsb = ZTOZSB(zp);
int count = 0;
sa_bulk_attr_t *bulk, *sa_attrs;
zfs_acl_locator_cb_t locate = { 0 };
uint64_t uid, gid, mode, rdev, xattr, parent;
uint64_t crtime[2], mtime[2], ctime[2];
zfs_acl_phys_t znode_acl;
char scanstamp[AV_SCANSTAMP_SZ];
boolean_t drop_lock = B_FALSE;
/*
* No upgrade if ACL isn't cached
* since we won't know which locks are held
* and ready the ACL would require special "locked"
* interfaces that would be messy
*/
if (zp->z_acl_cached == NULL || S_ISLNK(ZTOI(zp)->i_mode))
return;
/*
* If the z_lock is held and we aren't the owner
* the just return since we don't want to deadlock
* trying to update the status of z_is_sa. This
* file can then be upgraded at a later time.
*
* Otherwise, we know we are doing the
* sa_update() that caused us to enter this function.
*/
if (mutex_owner(&zp->z_lock) != curthread) {
if (mutex_tryenter(&zp->z_lock) == 0)
return;
else
drop_lock = B_TRUE;
}
/* First do a bulk query of the attributes that aren't cached */
bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * 20, KM_SLEEP);
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_CRTIME(zsb), NULL, &crtime, 16);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &mode, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zsb), NULL, &parent, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_XATTR(zsb), NULL, &xattr, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_RDEV(zsb), NULL, &rdev, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL, &uid, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb), NULL, &gid, 8);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ZNODE_ACL(zsb), NULL,
&znode_acl, 88);
if (sa_bulk_lookup_locked(hdl, bulk, count) != 0) {
kmem_free(bulk, sizeof (sa_bulk_attr_t) * 20);
goto done;
}
/*
* While the order here doesn't matter its best to try and organize
* it is such a way to pick up an already existing layout number
*/
count = 0;
sa_attrs = kmem_zalloc(sizeof (sa_bulk_attr_t) * 20, KM_SLEEP);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_MODE(zsb), NULL, &mode, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_SIZE(zsb), NULL,
&zp->z_size, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_GEN(zsb),
NULL, &zp->z_gen, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_UID(zsb), NULL, &uid, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_GID(zsb), NULL, &gid, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_PARENT(zsb),
NULL, &parent, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_FLAGS(zsb), NULL,
&zp->z_pflags, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_ATIME(zsb), NULL,
zp->z_atime, 16);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_MTIME(zsb), NULL,
&mtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_CTIME(zsb), NULL,
&ctime, 16);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_CRTIME(zsb), NULL,
&crtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_LINKS(zsb), NULL,
&zp->z_links, 8);
if (S_ISBLK(ZTOI(zp)->i_mode) || S_ISCHR(ZTOI(zp)->i_mode))
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_RDEV(zsb), NULL,
&rdev, 8);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_DACL_COUNT(zsb), NULL,
&zp->z_acl_cached->z_acl_count, 8);
if (zp->z_acl_cached->z_version < ZFS_ACL_VERSION_FUID)
zfs_acl_xform(zp, zp->z_acl_cached, CRED());
locate.cb_aclp = zp->z_acl_cached;
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_DACL_ACES(zsb),
zfs_acl_data_locator, &locate, zp->z_acl_cached->z_acl_bytes);
if (xattr)
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_XATTR(zsb),
NULL, &xattr, 8);
/* if scanstamp then add scanstamp */
if (zp->z_pflags & ZFS_BONUS_SCANSTAMP) {
abd_copy_to_buf_off(scanstamp, db->db_data,
AV_SCANSTAMP_SZ, ZFS_OLD_ZNODE_PHYS_SIZE);
SA_ADD_BULK_ATTR(sa_attrs, count, SA_ZPL_SCANSTAMP(zsb),
NULL, scanstamp, AV_SCANSTAMP_SZ);
zp->z_pflags &= ~ZFS_BONUS_SCANSTAMP;
}
VERIFY(dmu_set_bonustype(db, DMU_OT_SA, tx) == 0);
VERIFY(sa_replace_all_by_template_locked(hdl, sa_attrs,
count, tx) == 0);
if (znode_acl.z_acl_extern_obj)
VERIFY(0 == dmu_object_free(zsb->z_os,
znode_acl.z_acl_extern_obj, tx));
zp->z_is_sa = B_TRUE;
kmem_free(sa_attrs, sizeof (sa_bulk_attr_t) * 20);
kmem_free(bulk, sizeof (sa_bulk_attr_t) * 20);
done:
if (drop_lock)
mutex_exit(&zp->z_lock);
}
/*
* Lock a directory entry. A dirlock on <dzp, name> protects that name
* in dzp's directory zap object. As long as you hold a dirlock, you can
* assume two things: (1) dzp cannot be reaped, and (2) no other thread
* can change the zap entry for (i.e. link or unlink) this name.
*
* Input arguments:
* dzp - znode for directory
* name - name of entry to lock
* flag - ZNEW: if the entry already exists, fail with EEXIST.
* ZEXISTS: if the entry does not exist, fail with ENOENT.
* ZSHARED: allow concurrent access with other ZSHARED callers.
* ZXATTR: we want dzp's xattr directory
* ZCILOOK: On a mixed sensitivity file system,
* this lookup should be case-insensitive.
* ZCIEXACT: On a purely case-insensitive file system,
* this lookup should be case-sensitive.
* ZRENAMING: we are locking for renaming, force narrow locks
* ZHAVELOCK: Don't grab the z_name_lock for this call. The
* current thread already holds it.
*
* Output arguments:
* zpp - pointer to the znode for the entry (NULL if there isn't one)
* dlpp - pointer to the dirlock for this entry (NULL on error)
* direntflags - (case-insensitive lookup only)
* flags if multiple case-sensitive matches exist in directory
* realpnp - (case-insensitive lookup only)
* actual name matched within the directory
*
* Return value: 0 on success or errno on failure.
*
* NOTE: Always checks for, and rejects, '.' and '..'.
* NOTE: For case-insensitive file systems we take wide locks (see below),
* but return znode pointers to a single match.
*/
int
zfs_dirent_lock(zfs_dirlock_t **dlpp, znode_t *dzp, char *name, znode_t **zpp,
int flag, int *direntflags, pathname_t *realpnp)
{
zfs_sb_t *zsb = ZTOZSB(dzp);
zfs_dirlock_t *dl;
boolean_t update;
boolean_t exact;
uint64_t zoid;
#ifdef HAVE_DNLC
vnode_t *vp = NULL;
#endif /* HAVE_DNLC */
int error = 0;
int cmpflags;
*zpp = NULL;
*dlpp = NULL;
/*
* Verify that we are not trying to lock '.', '..', or '.zfs'
*/
if ((name[0] == '.' &&
(name[1] == '\0' || (name[1] == '.' && name[2] == '\0'))) ||
(zfs_has_ctldir(dzp) && strcmp(name, ZFS_CTLDIR_NAME) == 0))
return (EEXIST);
/*
* Case sensitivity and normalization preferences are set when
* the file system is created. These are stored in the
* zsb->z_case and zsb->z_norm fields. These choices
* affect what vnodes can be cached in the DNLC, how we
* perform zap lookups, and the "width" of our dirlocks.
*
* A normal dirlock locks a single name. Note that with
* normalization a name can be composed multiple ways, but
* when normalized, these names all compare equal. A wide
* dirlock locks multiple names. We need these when the file
* system is supporting mixed-mode access. It is sometimes
* necessary to lock all case permutations of file name at
* once so that simultaneous case-insensitive/case-sensitive
* behaves as rationally as possible.
*/
/*
* Decide if exact matches should be requested when performing
* a zap lookup on file systems supporting case-insensitive
* access.
*/
exact =
((zsb->z_case == ZFS_CASE_INSENSITIVE) && (flag & ZCIEXACT)) ||
((zsb->z_case == ZFS_CASE_MIXED) && !(flag & ZCILOOK));
/*
* Only look in or update the DNLC if we are looking for the
* name on a file system that does not require normalization
* or case folding. We can also look there if we happen to be
* on a non-normalizing, mixed sensitivity file system IF we
* are looking for the exact name.
*
* Maybe can add TO-UPPERed version of name to dnlc in ci-only
* case for performance improvement?
*/
update = !zsb->z_norm ||
((zsb->z_case == ZFS_CASE_MIXED) &&
!(zsb->z_norm & ~U8_TEXTPREP_TOUPPER) && !(flag & ZCILOOK));
/*
* ZRENAMING indicates we are in a situation where we should
* take narrow locks regardless of the file system's
* preferences for normalizing and case folding. This will
* prevent us deadlocking trying to grab the same wide lock
* twice if the two names happen to be case-insensitive
* matches.
*/
if (flag & ZRENAMING)
cmpflags = 0;
else
cmpflags = zsb->z_norm;
/*
* Wait until there are no locks on this name.
*
* Don't grab the the lock if it is already held. However, cannot
* have both ZSHARED and ZHAVELOCK together.
*/
ASSERT(!(flag & ZSHARED) || !(flag & ZHAVELOCK));
if (!(flag & ZHAVELOCK))
rw_enter(&dzp->z_name_lock, RW_READER);
mutex_enter(&dzp->z_lock);
for (;;) {
if (dzp->z_unlinked) {
mutex_exit(&dzp->z_lock);
if (!(flag & ZHAVELOCK))
rw_exit(&dzp->z_name_lock);
return (ENOENT);
}
for (dl = dzp->z_dirlocks; dl != NULL; dl = dl->dl_next) {
if ((u8_strcmp(name, dl->dl_name, 0, cmpflags,
U8_UNICODE_LATEST, &error) == 0) || error != 0)
break;
}
if (error != 0) {
mutex_exit(&dzp->z_lock);
if (!(flag & ZHAVELOCK))
rw_exit(&dzp->z_name_lock);
return (ENOENT);
}
if (dl == NULL) {
/*
* Allocate a new dirlock and add it to the list.
*/
dl = kmem_alloc(sizeof (zfs_dirlock_t), KM_SLEEP);
cv_init(&dl->dl_cv, NULL, CV_DEFAULT, NULL);
dl->dl_name = name;
dl->dl_sharecnt = 0;
dl->dl_namelock = 0;
dl->dl_namesize = 0;
dl->dl_dzp = dzp;
dl->dl_next = dzp->z_dirlocks;
dzp->z_dirlocks = dl;
break;
}
if ((flag & ZSHARED) && dl->dl_sharecnt != 0)
break;
cv_wait(&dl->dl_cv, &dzp->z_lock);
}
/*
* If the z_name_lock was NOT held for this dirlock record it.
*/
if (flag & ZHAVELOCK)
dl->dl_namelock = 1;
if ((flag & ZSHARED) && ++dl->dl_sharecnt > 1 && dl->dl_namesize == 0) {
/*
* We're the second shared reference to dl. Make a copy of
* dl_name in case the first thread goes away before we do.
* Note that we initialize the new name before storing its
* pointer into dl_name, because the first thread may load
* dl->dl_name at any time. He'll either see the old value,
* which is his, or the new shared copy; either is OK.
*/
dl->dl_namesize = strlen(dl->dl_name) + 1;
name = kmem_alloc(dl->dl_namesize, KM_SLEEP);
bcopy(dl->dl_name, name, dl->dl_namesize);
dl->dl_name = name;
}
mutex_exit(&dzp->z_lock);
/*
* We have a dirlock on the name. (Note that it is the dirlock,
* not the dzp's z_lock, that protects the name in the zap object.)
* See if there's an object by this name; if so, put a hold on it.
*/
if (flag & ZXATTR) {
error = sa_lookup(dzp->z_sa_hdl, SA_ZPL_XATTR(zsb), &zoid,
sizeof (zoid));
if (error == 0)
error = (zoid == 0 ? ENOENT : 0);
} else {
#ifdef HAVE_DNLC
if (update)
vp = dnlc_lookup(ZTOI(dzp), name);
if (vp == DNLC_NO_VNODE) {
iput(vp);
error = ENOENT;
} else if (vp) {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
iput(vp);
return (EEXIST);
}
*dlpp = dl;
*zpp = VTOZ(vp);
return (0);
} else {
error = zfs_match_find(zsb, dzp, name, exact,
update, direntflags, realpnp, &zoid);
}
#else
error = zfs_match_find(zsb, dzp, name, exact,
update, direntflags, realpnp, &zoid);
#endif /* HAVE_DNLC */
}
if (error) {
if (error != ENOENT || (flag & ZEXISTS)) {
zfs_dirent_unlock(dl);
return (error);
}
} else {
if (flag & ZNEW) {
zfs_dirent_unlock(dl);
return (EEXIST);
}
error = zfs_zget(zsb, zoid, zpp);
if (error) {
zfs_dirent_unlock(dl);
return (error);
}
#ifdef HAVE_DNLC
if (!(flag & ZXATTR) && update)
dnlc_update(ZTOI(dzp), name, ZTOI(*zpp));
#endif /* HAVE_DNLC */
}
*dlpp = dl;
return (0);
}
/*
* Create a new DMU object to hold a zfs znode.
*
* IN: dzp - parent directory for new znode
* vap - file attributes for new znode
* tx - dmu transaction id for zap operations
* cr - credentials of caller
* flag - flags:
* IS_ROOT_NODE - new object will be root
* IS_XATTR - new object is an attribute
* bonuslen - length of bonus buffer
* setaclp - File/Dir initial ACL
* fuidp - Tracks fuid allocation.
*
* OUT: zpp - allocated znode
*
*/
void
zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr,
uint_t flag, znode_t **zpp, zfs_acl_ids_t *acl_ids)
{
uint64_t crtime[2], atime[2], mtime[2], ctime[2];
uint64_t mode, size, links, parent, pflags;
uint64_t dzp_pflags = 0;
uint64_t rdev = 0;
zfs_sb_t *zsb = ZTOZSB(dzp);
dmu_buf_t *db;
timestruc_t now;
uint64_t gen, obj;
int err;
int bonuslen;
sa_handle_t *sa_hdl;
dmu_object_type_t obj_type;
sa_bulk_attr_t *sa_attrs;
int cnt = 0;
zfs_acl_locator_cb_t locate = { 0 };
if (zsb->z_replay) {
obj = vap->va_nodeid;
now = vap->va_ctime; /* see zfs_replay_create() */
gen = vap->va_nblocks; /* ditto */
} else {
obj = 0;
gethrestime(&now);
gen = dmu_tx_get_txg(tx);
}
obj_type = zsb->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE;
bonuslen = (obj_type == DMU_OT_SA) ?
DN_MAX_BONUSLEN : ZFS_OLD_ZNODE_PHYS_SIZE;
/*
* Create a new DMU object.
*/
/*
* There's currently no mechanism for pre-reading the blocks that will
* be needed to allocate a new object, so we accept the small chance
* that there will be an i/o error and we will fail one of the
* assertions below.
*/
if (S_ISDIR(vap->va_mode)) {
if (zsb->z_replay) {
err = zap_create_claim_norm(zsb->z_os, obj,
zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS,
obj_type, bonuslen, tx);
ASSERT0(err);
} else {
obj = zap_create_norm(zsb->z_os,
zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS,
obj_type, bonuslen, tx);
}
} else {
if (zsb->z_replay) {
err = dmu_object_claim(zsb->z_os, obj,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
obj_type, bonuslen, tx);
ASSERT0(err);
} else {
obj = dmu_object_alloc(zsb->z_os,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
obj_type, bonuslen, tx);
}
}
ZFS_OBJ_HOLD_ENTER(zsb, obj);
VERIFY(0 == sa_buf_hold(zsb->z_os, obj, NULL, &db));
/*
* If this is the root, fix up the half-initialized parent pointer
* to reference the just-allocated physical data area.
*/
if (flag & IS_ROOT_NODE) {
dzp->z_id = obj;
} else {
dzp_pflags = dzp->z_pflags;
}
/*
* If parent is an xattr, so am I.
*/
if (dzp_pflags & ZFS_XATTR) {
flag |= IS_XATTR;
}
if (zsb->z_use_fuids)
pflags = ZFS_ARCHIVE | ZFS_AV_MODIFIED;
else
pflags = 0;
if (S_ISDIR(vap->va_mode)) {
size = 2; /* contents ("." and "..") */
links = (flag & (IS_ROOT_NODE | IS_XATTR)) ? 2 : 1;
} else {
size = links = 0;
}
if (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))
rdev = vap->va_rdev;
parent = dzp->z_id;
mode = acl_ids->z_mode;
if (flag & IS_XATTR)
pflags |= ZFS_XATTR;
/*
* No execs denied will be deterimed when zfs_mode_compute() is called.
*/
pflags |= acl_ids->z_aclp->z_hints &
(ZFS_ACL_TRIVIAL|ZFS_INHERIT_ACE|ZFS_ACL_AUTO_INHERIT|
ZFS_ACL_DEFAULTED|ZFS_ACL_PROTECTED);
ZFS_TIME_ENCODE(&now, crtime);
ZFS_TIME_ENCODE(&now, ctime);
if (vap->va_mask & ATTR_ATIME) {
ZFS_TIME_ENCODE(&vap->va_atime, atime);
} else {
ZFS_TIME_ENCODE(&now, atime);
}
if (vap->va_mask & ATTR_MTIME) {
ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
} else {
ZFS_TIME_ENCODE(&now, mtime);
}
/* Now add in all of the "SA" attributes */
VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, NULL, SA_HDL_SHARED,
&sa_hdl));
/*
* Setup the array of attributes to be replaced/set on the new file
*
* order for DMU_OT_ZNODE is critical since it needs to be constructed
* in the old znode_phys_t format. Don't change this ordering
*/
sa_attrs = kmem_alloc(sizeof (sa_bulk_attr_t) * ZPL_END, KM_PUSHPAGE);
if (obj_type == DMU_OT_ZNODE) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb),
NULL, &atime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb),
NULL, &mtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb),
NULL, &ctime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb),
NULL, &crtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb),
NULL, &gen, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb),
NULL, &mode, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb),
NULL, &size, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb),
NULL, &parent, 8);
} else {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb),
NULL, &mode, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb),
NULL, &size, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb),
NULL, &gen, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb),
NULL, &acl_ids->z_fuid, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb),
NULL, &acl_ids->z_fgid, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb),
NULL, &parent, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb),
NULL, &pflags, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb),
NULL, &atime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb),
NULL, &mtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb),
NULL, &ctime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb),
NULL, &crtime, 16);
}
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zsb), NULL, &links, 8);
if (obj_type == DMU_OT_ZNODE) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zsb), NULL,
&empty_xattr, 8);
}
if (obj_type == DMU_OT_ZNODE ||
(S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zsb),
NULL, &rdev, 8);
}
if (obj_type == DMU_OT_ZNODE) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb),
NULL, &pflags, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), NULL,
&acl_ids->z_fuid, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb), NULL,
&acl_ids->z_fgid, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PAD(zsb), NULL, pad,
sizeof (uint64_t) * 4);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zsb), NULL,
&acl_phys, sizeof (zfs_acl_phys_t));
} else if (acl_ids->z_aclp->z_version >= ZFS_ACL_VERSION_FUID) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_COUNT(zsb), NULL,
&acl_ids->z_aclp->z_acl_count, 8);
locate.cb_aclp = acl_ids->z_aclp;
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_ACES(zsb),
zfs_acl_data_locator, &locate,
acl_ids->z_aclp->z_acl_bytes);
mode = zfs_mode_compute(mode, acl_ids->z_aclp, &pflags,
acl_ids->z_fuid, acl_ids->z_fgid);
}
VERIFY(sa_replace_all_by_template(sa_hdl, sa_attrs, cnt, tx) == 0);
if (!(flag & IS_ROOT_NODE)) {
*zpp = zfs_znode_alloc(zsb, db, 0, obj_type, obj, sa_hdl,
ZTOI(dzp));
VERIFY(*zpp != NULL);
VERIFY(dzp != NULL);
} else {
/*
* If we are creating the root node, the "parent" we
* passed in is the znode for the root.
*/
*zpp = dzp;
(*zpp)->z_sa_hdl = sa_hdl;
}
(*zpp)->z_pflags = pflags;
(*zpp)->z_mode = mode;
if (obj_type == DMU_OT_ZNODE ||
acl_ids->z_aclp->z_version < ZFS_ACL_VERSION_FUID) {
err = zfs_aclset_common(*zpp, acl_ids->z_aclp, cr, tx);
ASSERT0(err);
}
kmem_free(sa_attrs, sizeof (sa_bulk_attr_t) * ZPL_END);
ZFS_OBJ_HOLD_EXIT(zsb, obj);
}
/*
* 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);
}
int
zfs_vget(struct super_block *sb, struct inode **ipp, fid_t *fidp)
{
zfs_sb_t *zsb = sb->s_fs_info;
znode_t *zp;
uint64_t object = 0;
uint64_t fid_gen = 0;
uint64_t gen_mask;
uint64_t zp_gen;
int i, err;
*ipp = NULL;
ZFS_ENTER(zsb);
if (fidp->fid_len == LONG_FID_LEN) {
zfid_long_t *zlfid = (zfid_long_t *)fidp;
uint64_t objsetid = 0;
uint64_t setgen = 0;
for (i = 0; i < sizeof (zlfid->zf_setid); i++)
objsetid |= ((uint64_t)zlfid->zf_setid[i]) << (8 * i);
for (i = 0; i < sizeof (zlfid->zf_setgen); i++)
setgen |= ((uint64_t)zlfid->zf_setgen[i]) << (8 * i);
ZFS_EXIT(zsb);
err = zfsctl_lookup_objset(sb, objsetid, &zsb);
if (err)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zsb);
}
if (fidp->fid_len == SHORT_FID_LEN || fidp->fid_len == LONG_FID_LEN) {
zfid_short_t *zfid = (zfid_short_t *)fidp;
for (i = 0; i < sizeof (zfid->zf_object); i++)
object |= ((uint64_t)zfid->zf_object[i]) << (8 * i);
for (i = 0; i < sizeof (zfid->zf_gen); i++)
fid_gen |= ((uint64_t)zfid->zf_gen[i]) << (8 * i);
} else {
ZFS_EXIT(zsb);
return (SET_ERROR(EINVAL));
}
/* A zero fid_gen means we are in the .zfs control directories */
if (fid_gen == 0 &&
(object == ZFSCTL_INO_ROOT || object == ZFSCTL_INO_SNAPDIR)) {
*ipp = zsb->z_ctldir;
ASSERT(*ipp != NULL);
if (object == ZFSCTL_INO_SNAPDIR) {
VERIFY(zfsctl_root_lookup(*ipp, "snapshot", ipp,
0, kcred, NULL, NULL) == 0);
} else {
igrab(*ipp);
}
ZFS_EXIT(zsb);
return (0);
}
gen_mask = -1ULL >> (64 - 8 * i);
dprintf("getting %llu [%llu mask %llx]\n", object, fid_gen, gen_mask);
if ((err = zfs_zget(zsb, object, &zp))) {
ZFS_EXIT(zsb);
return (err);
}
(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zsb), &zp_gen,
sizeof (uint64_t));
zp_gen = zp_gen & gen_mask;
if (zp_gen == 0)
zp_gen = 1;
if (zp->z_unlinked || zp_gen != fid_gen) {
dprintf("znode gen (%llu) != fid gen (%llu)\n", zp_gen,
fid_gen);
iput(ZTOI(zp));
ZFS_EXIT(zsb);
return (SET_ERROR(EINVAL));
}
*ipp = ZTOI(zp);
if (*ipp)
zfs_inode_update(ITOZ(*ipp));
ZFS_EXIT(zsb);
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);
}
/*
* Handles TX_CREATE, TX_CREATE_ATTR, TX_MKDIR, TX_MKDIR_ATTR and
* TK_MKXATTR transactions.
*
* TX_CREATE and TX_MKDIR are standard creates, but they may have FUID
* domain information appended prior to the name. In this case the
* uid/gid in the log record will be a log centric FUID.
*
* TX_CREATE_ACL_ATTR and TX_MKDIR_ACL_ATTR handle special creates that
* may contain attributes, ACL and optional fuid information.
*
* TX_CREATE_ACL and TX_MKDIR_ACL handle special creates that specify
* and ACL and normal users/groups in the ACEs.
*
* There may be an optional xvattr attribute information similar
* to zfs_log_setattr.
*
* Also, after the file name "domain" strings may be appended.
*/
void
zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
znode_t *dzp, znode_t *zp, char *name, vsecattr_t *vsecp,
zfs_fuid_info_t *fuidp, vattr_t *vap)
{
itx_t *itx;
lr_create_t *lr;
lr_acl_create_t *lracl;
size_t aclsize = 0;
size_t xvatsize = 0;
size_t txsize;
xvattr_t *xvap = (xvattr_t *)vap;
void *end;
size_t lrsize;
size_t namesize = strlen(name) + 1;
size_t fuidsz = 0;
if (zil_replaying(zilog, tx) || zfs_xattr_owner_unlinked(dzp))
return;
/*
* If we have FUIDs present then add in space for
* domains and ACE fuid's if any.
*/
if (fuidp) {
fuidsz += fuidp->z_domain_str_sz;
fuidsz += fuidp->z_fuid_cnt * sizeof (uint64_t);
}
if (vap->va_mask & ATTR_XVATTR)
xvatsize = ZIL_XVAT_SIZE(xvap->xva_mapsize);
if ((int)txtype == TX_CREATE_ATTR || (int)txtype == TX_MKDIR_ATTR ||
(int)txtype == TX_CREATE || (int)txtype == TX_MKDIR ||
(int)txtype == TX_MKXATTR) {
txsize = sizeof (*lr) + namesize + fuidsz + xvatsize;
lrsize = sizeof (*lr);
} else {
txsize =
sizeof (lr_acl_create_t) + namesize + fuidsz +
ZIL_ACE_LENGTH(aclsize) + xvatsize;
lrsize = sizeof (lr_acl_create_t);
}
itx = zil_itx_create(txtype, txsize);
lr = (lr_create_t *)&itx->itx_lr;
lr->lr_doid = dzp->z_id;
lr->lr_foid = zp->z_id;
/* Store dnode slot count in 8 bits above object id. */
LR_FOID_SET_SLOTS(lr->lr_foid, zp->z_dnodesize >> DNODE_SHIFT);
lr->lr_mode = zp->z_mode;
if (!IS_EPHEMERAL(KUID_TO_SUID(ZTOI(zp)->i_uid))) {
lr->lr_uid = (uint64_t)KUID_TO_SUID(ZTOI(zp)->i_uid);
} else {
lr->lr_uid = fuidp->z_fuid_owner;
}
if (!IS_EPHEMERAL(KGID_TO_SGID(ZTOI(zp)->i_gid))) {
lr->lr_gid = (uint64_t)KGID_TO_SGID(ZTOI(zp)->i_gid);
} else {
lr->lr_gid = fuidp->z_fuid_group;
}
(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(ZTOZSB(zp)), &lr->lr_gen,
sizeof (uint64_t));
(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(ZTOZSB(zp)),
lr->lr_crtime, sizeof (uint64_t) * 2);
if (sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(ZTOZSB(zp)), &lr->lr_rdev,
sizeof (lr->lr_rdev)) != 0)
lr->lr_rdev = 0;
/*
* Fill in xvattr info if any
*/
if (vap->va_mask & ATTR_XVATTR) {
zfs_log_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), xvap);
end = (caddr_t)lr + lrsize + xvatsize;
} else {
end = (caddr_t)lr + lrsize;
}
/* Now fill in any ACL info */
if (vsecp) {
lracl = (lr_acl_create_t *)&itx->itx_lr;
lracl->lr_aclcnt = vsecp->vsa_aclcnt;
lracl->lr_acl_bytes = aclsize;
lracl->lr_domcnt = fuidp ? fuidp->z_domain_cnt : 0;
lracl->lr_fuidcnt = fuidp ? fuidp->z_fuid_cnt : 0;
if (vsecp->vsa_aclflags & VSA_ACE_ACLFLAGS)
lracl->lr_acl_flags = (uint64_t)vsecp->vsa_aclflags;
else
lracl->lr_acl_flags = 0;
bcopy(vsecp->vsa_aclentp, end, aclsize);
end = (caddr_t)end + ZIL_ACE_LENGTH(aclsize);
}
/* drop in FUID info */
if (fuidp) {
end = zfs_log_fuid_ids(fuidp, end);
end = zfs_log_fuid_domains(fuidp, end);
}
/*
* Now place file name in log record
*/
bcopy(name, end, namesize);
zil_itx_assign(zilog, itx, tx);
}
int
zfs_zget(zfs_sb_t *zsb, uint64_t obj_num, znode_t **zpp)
{
dmu_object_info_t doi;
dmu_buf_t *db;
znode_t *zp;
int err;
sa_handle_t *hdl;
*zpp = NULL;
again:
ZFS_OBJ_HOLD_ENTER(zsb, obj_num);
err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db);
if (err) {
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
return (err);
}
dmu_object_info_from_db(db, &doi);
if (doi.doi_bonus_type != DMU_OT_SA &&
(doi.doi_bonus_type != DMU_OT_ZNODE ||
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
return (SET_ERROR(EINVAL));
}
hdl = dmu_buf_get_user(db);
if (hdl != NULL) {
zp = sa_get_userdata(hdl);
/*
* Since "SA" does immediate eviction we
* should never find a sa handle that doesn't
* know about the znode.
*/
ASSERT3P(zp, !=, NULL);
mutex_enter(&zp->z_lock);
ASSERT3U(zp->z_id, ==, obj_num);
if (zp->z_unlinked) {
err = SET_ERROR(ENOENT);
} else {
/*
* If igrab() returns NULL the VFS has independently
* determined the inode should be evicted and has
* called iput_final() to start the eviction process.
* The SA handle is still valid but because the VFS
* requires that the eviction succeed we must drop
* our locks and references to allow the eviction to
* complete. The zfs_zget() may then be retried.
*
* This unlikely case could be optimized by registering
* a sops->drop_inode() callback. The callback would
* need to detect the active SA hold thereby informing
* the VFS that this inode should not be evicted.
*/
if (igrab(ZTOI(zp)) == NULL) {
mutex_exit(&zp->z_lock);
sa_buf_rele(db, NULL);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
goto again;
}
*zpp = zp;
err = 0;
}
mutex_exit(&zp->z_lock);
sa_buf_rele(db, NULL);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
return (err);
}
static int
zfs_replay_write(zfs_sb_t *zsb, lr_write_t *lr, boolean_t byteswap)
{
char *data = (char *)(lr + 1); /* data follows lr_write_t */
znode_t *zp;
int error;
uint64_t eod, offset, length;
if (byteswap)
byteswap_uint64_array(lr, sizeof (*lr));
if ((error = zfs_zget(zsb, lr->lr_foid, &zp)) != 0) {
/*
* As we can log writes out of order, it's possible the
* file has been removed. In this case just drop the write
* and return success.
*/
if (error == ENOENT)
error = 0;
return (error);
}
offset = lr->lr_offset;
length = lr->lr_length;
eod = offset + length; /* end of data for this write */
/*
* This may be a write from a dmu_sync() for a whole block,
* and may extend beyond the current end of the file.
* We can't just replay what was written for this TX_WRITE as
* a future TX_WRITE2 may extend the eof and the data for that
* write needs to be there. So we write the whole block and
* reduce the eof. This needs to be done within the single dmu
* transaction created within vn_rdwr -> zfs_write. So a possible
* new end of file is passed through in zsb->z_replay_eof
*/
zsb->z_replay_eof = 0; /* 0 means don't change end of file */
/* If it's a dmu_sync() block, write the whole block */
if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) {
uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr);
if (length < blocksize) {
offset -= offset % blocksize;
length = blocksize;
}
if (zp->z_size < eod)
zsb->z_replay_eof = eod;
}
error = zpl_write_common(ZTOI(zp), data, length, offset,
UIO_SYSSPACE, 0, kcred);
if (error) {
if (error < 0)
error = -error;
else
error = EIO; /* Short write */
}
iput(ZTOI(zp));
zsb->z_replay_eof = 0; /* safety */
return (error);
}
static int
zfs_replay_create(zfs_sb_t *zsb, lr_create_t *lr, boolean_t byteswap)
{
char *name = NULL; /* location determined later */
char *link; /* symlink content follows name */
znode_t *dzp;
struct inode *ip = NULL;
xvattr_t xva;
int vflg = 0;
size_t lrsize = sizeof (lr_create_t);
lr_attr_t *lrattr;
void *start;
size_t xvatlen;
uint64_t txtype;
int error;
txtype = (lr->lr_common.lrc_txtype & ~TX_CI);
if (byteswap) {
byteswap_uint64_array(lr, sizeof (*lr));
if (txtype == TX_CREATE_ATTR || txtype == TX_MKDIR_ATTR)
zfs_replay_swap_attrs((lr_attr_t *)(lr + 1));
}
if ((error = zfs_zget(zsb, lr->lr_doid, &dzp)) != 0)
return (error);
xva_init(&xva);
zfs_init_vattr(&xva.xva_vattr, ATTR_MODE | ATTR_UID | ATTR_GID,
lr->lr_mode, lr->lr_uid, lr->lr_gid, lr->lr_rdev, lr->lr_foid);
/*
* All forms of zfs create (create, mkdir, mkxattrdir, symlink)
* eventually end up in zfs_mknode(), which assigns the object's
* creation time and generation number. The generic zfs_create()
* doesn't have either concept, so we smuggle the values inside
* the vattr's otherwise unused va_ctime and va_nblocks fields.
*/
ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
xva.xva_vattr.va_nblocks = lr->lr_gen;
error = dmu_object_info(zsb->z_os, lr->lr_foid, NULL);
if (error != ENOENT)
goto out;
if (lr->lr_common.lrc_txtype & TX_CI)
vflg |= FIGNORECASE;
/*
* Symlinks don't have fuid info, and CIFS never creates
* symlinks.
*
* The _ATTR versions will grab the fuid info in their subcases.
*/
if ((int)lr->lr_common.lrc_txtype != TX_SYMLINK &&
(int)lr->lr_common.lrc_txtype != TX_MKDIR_ATTR &&
(int)lr->lr_common.lrc_txtype != TX_CREATE_ATTR) {
start = (lr + 1);
zsb->z_fuid_replay =
zfs_replay_fuid_domain(start, &start,
lr->lr_uid, lr->lr_gid);
}
switch (txtype) {
case TX_CREATE_ATTR:
lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
start = (caddr_t)(lr + 1) + xvatlen;
zsb->z_fuid_replay =
zfs_replay_fuid_domain(start, &start,
lr->lr_uid, lr->lr_gid);
name = (char *)start;
/*FALLTHROUGH*/
case TX_CREATE:
if (name == NULL)
name = (char *)start;
error = zfs_create(ZTOI(dzp), name, &xva.xva_vattr,
0, 0, &ip, kcred, vflg, NULL);
break;
case TX_MKDIR_ATTR:
lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
start = (caddr_t)(lr + 1) + xvatlen;
zsb->z_fuid_replay =
zfs_replay_fuid_domain(start, &start,
lr->lr_uid, lr->lr_gid);
name = (char *)start;
/*FALLTHROUGH*/
case TX_MKDIR:
if (name == NULL)
name = (char *)(lr + 1);
error = zfs_mkdir(ZTOI(dzp), name, &xva.xva_vattr,
&ip, kcred, vflg, NULL);
break;
case TX_MKXATTR:
error = zfs_make_xattrdir(dzp, &xva.xva_vattr, &ip, kcred);
break;
case TX_SYMLINK:
name = (char *)(lr + 1);
link = name + strlen(name) + 1;
error = zfs_symlink(ZTOI(dzp), name, &xva.xva_vattr,
link, &ip, kcred, vflg);
break;
default:
error = ENOTSUP;
}
out:
if (error == 0 && ip != NULL)
iput(ip);
iput(ZTOI(dzp));
if (zsb->z_fuid_replay)
zfs_fuid_info_free(zsb->z_fuid_replay);
zsb->z_fuid_replay = NULL;
return (error);
}
/*
* Replay file create with optional ACL, xvattr information as well
* as option FUID information.
*/
static int
zfs_replay_create_acl(zfs_sb_t *zsb, lr_acl_create_t *lracl, boolean_t byteswap)
{
char *name = NULL; /* location determined later */
lr_create_t *lr = (lr_create_t *)lracl;
znode_t *dzp;
struct inode *ip = NULL;
xvattr_t xva;
int vflg = 0;
vsecattr_t vsec = { 0 };
lr_attr_t *lrattr;
void *aclstart;
void *fuidstart;
size_t xvatlen = 0;
uint64_t txtype;
int error;
txtype = (lr->lr_common.lrc_txtype & ~TX_CI);
if (byteswap) {
byteswap_uint64_array(lracl, sizeof (*lracl));
if (txtype == TX_CREATE_ACL_ATTR ||
txtype == TX_MKDIR_ACL_ATTR) {
lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
zfs_replay_swap_attrs(lrattr);
xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
}
aclstart = (caddr_t)(lracl + 1) + xvatlen;
zfs_ace_byteswap(aclstart, lracl->lr_acl_bytes, B_FALSE);
/* swap fuids */
if (lracl->lr_fuidcnt) {
byteswap_uint64_array((caddr_t)aclstart +
ZIL_ACE_LENGTH(lracl->lr_acl_bytes),
lracl->lr_fuidcnt * sizeof (uint64_t));
}
}
if ((error = zfs_zget(zsb, lr->lr_doid, &dzp)) != 0)
return (error);
xva_init(&xva);
zfs_init_vattr(&xva.xva_vattr, ATTR_MODE | ATTR_UID | ATTR_GID,
lr->lr_mode, lr->lr_uid, lr->lr_gid, lr->lr_rdev, lr->lr_foid);
/*
* All forms of zfs create (create, mkdir, mkxattrdir, symlink)
* eventually end up in zfs_mknode(), which assigns the object's
* creation time and generation number. The generic zfs_create()
* doesn't have either concept, so we smuggle the values inside
* the vattr's otherwise unused va_ctime and va_nblocks fields.
*/
ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
xva.xva_vattr.va_nblocks = lr->lr_gen;
error = dmu_object_info(zsb->z_os, lr->lr_foid, NULL);
if (error != ENOENT)
goto bail;
if (lr->lr_common.lrc_txtype & TX_CI)
vflg |= FIGNORECASE;
switch (txtype) {
case TX_CREATE_ACL:
aclstart = (caddr_t)(lracl + 1);
fuidstart = (caddr_t)aclstart +
ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
zsb->z_fuid_replay = zfs_replay_fuids(fuidstart,
(void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
lr->lr_uid, lr->lr_gid);
/*FALLTHROUGH*/
case TX_CREATE_ACL_ATTR:
if (name == NULL) {
lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
xva.xva_vattr.va_mask |= ATTR_XVATTR;
zfs_replay_xvattr(lrattr, &xva);
}
vsec.vsa_mask = VSA_ACE | VSA_ACE_ACLFLAGS;
vsec.vsa_aclentp = (caddr_t)(lracl + 1) + xvatlen;
vsec.vsa_aclcnt = lracl->lr_aclcnt;
vsec.vsa_aclentsz = lracl->lr_acl_bytes;
vsec.vsa_aclflags = lracl->lr_acl_flags;
if (zsb->z_fuid_replay == NULL) {
fuidstart = (caddr_t)(lracl + 1) + xvatlen +
ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
zsb->z_fuid_replay =
zfs_replay_fuids(fuidstart,
(void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
lr->lr_uid, lr->lr_gid);
}
error = zfs_create(ZTOI(dzp), name, &xva.xva_vattr,
0, 0, &ip, kcred, vflg, &vsec);
break;
case TX_MKDIR_ACL:
aclstart = (caddr_t)(lracl + 1);
fuidstart = (caddr_t)aclstart +
ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
zsb->z_fuid_replay = zfs_replay_fuids(fuidstart,
(void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
lr->lr_uid, lr->lr_gid);
/*FALLTHROUGH*/
case TX_MKDIR_ACL_ATTR:
if (name == NULL) {
lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
zfs_replay_xvattr(lrattr, &xva);
}
vsec.vsa_mask = VSA_ACE | VSA_ACE_ACLFLAGS;
vsec.vsa_aclentp = (caddr_t)(lracl + 1) + xvatlen;
vsec.vsa_aclcnt = lracl->lr_aclcnt;
vsec.vsa_aclentsz = lracl->lr_acl_bytes;
vsec.vsa_aclflags = lracl->lr_acl_flags;
if (zsb->z_fuid_replay == NULL) {
fuidstart = (caddr_t)(lracl + 1) + xvatlen +
ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
zsb->z_fuid_replay =
zfs_replay_fuids(fuidstart,
(void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
lr->lr_uid, lr->lr_gid);
}
error = zfs_mkdir(ZTOI(dzp), name, &xva.xva_vattr,
&ip, kcred, vflg, &vsec);
break;
default:
error = ENOTSUP;
}
bail:
if (error == 0 && ip != NULL)
iput(ip);
iput(ZTOI(dzp));
if (zsb->z_fuid_replay)
zfs_fuid_info_free(zsb->z_fuid_replay);
zsb->z_fuid_replay = NULL;
return (error);
}
/*
* Unlink zp from dl, and mark zp for deletion if this was the last link.
* Can fail if zp is a mount point (EBUSY) or a non-empty directory (EEXIST).
* If 'unlinkedp' is NULL, we put unlinked znodes on the unlinked list.
* If it's non-NULL, we use it to indicate whether the znode needs deletion,
* and it's the caller's job to do it.
*/
int
zfs_link_destroy(zfs_dirlock_t *dl, znode_t *zp, dmu_tx_t *tx, int flag,
boolean_t *unlinkedp)
{
znode_t *dzp = dl->dl_dzp;
zfs_sb_t *zsb = ZTOZSB(dzp);
int zp_is_dir = S_ISDIR(ZTOI(zp)->i_mode);
boolean_t unlinked = B_FALSE;
sa_bulk_attr_t bulk[5];
uint64_t mtime[2], ctime[2];
int count = 0;
int error;
#ifdef HAVE_DNLC
dnlc_remove(ZTOI(dzp), dl->dl_name);
#endif /* HAVE_DNLC */
if (!(flag & ZRENAMING)) {
mutex_enter(&zp->z_lock);
if (zp_is_dir && !zfs_dirempty(zp)) {
mutex_exit(&zp->z_lock);
return (EEXIST);
}
/*
* If we get here, we are going to try to remove the object.
* First try removing the name from the directory; if that
* fails, return the error.
*/
error = zfs_dropname(dl, zp, dzp, tx, flag);
if (error != 0) {
mutex_exit(&zp->z_lock);
return (error);
}
if (zp->z_links <= zp_is_dir) {
zfs_panic_recover("zfs: link count on %lu is %u, "
"should be at least %u", zp->z_id,
(int)zp->z_links, zp_is_dir + 1);
zp->z_links = zp_is_dir + 1;
}
if (--zp->z_links == zp_is_dir) {
zp->z_unlinked = B_TRUE;
zp->z_links = 0;
unlinked = B_TRUE;
} else {
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb),
NULL, &ctime, sizeof (ctime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb),
NULL, &zp->z_pflags, sizeof (zp->z_pflags));
zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime, ctime,
B_TRUE);
}
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb),
NULL, &zp->z_links, sizeof (zp->z_links));
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
count = 0;
ASSERT(error == 0);
mutex_exit(&zp->z_lock);
} else {
error = zfs_dropname(dl, zp, dzp, tx, flag);
if (error != 0)
return (error);
}
mutex_enter(&dzp->z_lock);
dzp->z_size--; /* one dirent removed */
dzp->z_links -= zp_is_dir; /* ".." link from zp */
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb),
NULL, &dzp->z_links, sizeof (dzp->z_links));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb),
NULL, &dzp->z_size, sizeof (dzp->z_size));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb),
NULL, ctime, sizeof (ctime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb),
NULL, mtime, sizeof (mtime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb),
NULL, &dzp->z_pflags, sizeof (dzp->z_pflags));
zfs_tstamp_update_setup(dzp, CONTENT_MODIFIED, mtime, ctime, B_TRUE);
error = sa_bulk_update(dzp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
mutex_exit(&dzp->z_lock);
if (unlinkedp != NULL)
*unlinkedp = unlinked;
else if (unlinked)
zfs_unlinked_add(zp, tx);
return (0);
}
/*
* 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);
}
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 (EDQUOT);
}
top:
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_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(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));
(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 = 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));
}
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));
}
/*
* Link zp into dl. Can only fail if zp has been unlinked.
*/
int
zfs_link_create(zfs_dirlock_t *dl, znode_t *zp, dmu_tx_t *tx, int flag)
{
znode_t *dzp = dl->dl_dzp;
zfsvfs_t *zfsvfs = ZTOZSB(zp);
uint64_t value;
int zp_is_dir = S_ISDIR(ZTOI(zp)->i_mode);
sa_bulk_attr_t bulk[5];
uint64_t mtime[2], ctime[2];
uint64_t links;
int count = 0;
int error;
mutex_enter(&zp->z_lock);
if (!(flag & ZRENAMING)) {
if (zp->z_unlinked) { /* no new links to unlinked zp */
ASSERT(!(flag & (ZNEW | ZEXISTS)));
mutex_exit(&zp->z_lock);
return (SET_ERROR(ENOENT));
}
if (!(flag & ZNEW)) {
/*
* ZNEW nodes come from zfs_mknode() where the link
* count has already been initialised
*/
inc_nlink(ZTOI(zp));
links = ZTOI(zp)->i_nlink;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs),
NULL, &links, sizeof (links));
}
}
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL,
&dzp->z_id, sizeof (dzp->z_id));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, sizeof (zp->z_pflags));
if (!(flag & ZNEW)) {
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
ctime, sizeof (ctime));
zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime,
ctime);
}
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
mutex_exit(&zp->z_lock);
mutex_enter(&dzp->z_lock);
dzp->z_size++;
if (zp_is_dir)
inc_nlink(ZTOI(dzp));
links = ZTOI(dzp)->i_nlink;
count = 0;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL,
&dzp->z_size, sizeof (dzp->z_size));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs), NULL,
&links, sizeof (links));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
mtime, sizeof (mtime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
ctime, sizeof (ctime));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&dzp->z_pflags, sizeof (dzp->z_pflags));
zfs_tstamp_update_setup(dzp, CONTENT_MODIFIED, mtime, ctime);
error = sa_bulk_update(dzp->z_sa_hdl, bulk, count, tx);
ASSERT(error == 0);
mutex_exit(&dzp->z_lock);
value = zfs_dirent(zp, zp->z_mode);
error = zap_add(ZTOZSB(zp)->z_os, dzp->z_id, dl->dl_name,
8, 1, &value, tx);
ASSERT(error == 0);
return (0);
}
int
zfs_zget(zfs_sb_t *zsb, uint64_t obj_num, znode_t **zpp)
{
dmu_object_info_t doi;
dmu_buf_t *db;
znode_t *zp;
int err;
sa_handle_t *hdl;
struct inode *ip;
*zpp = NULL;
again:
ip = ilookup(zsb->z_sb, obj_num);
ZFS_OBJ_HOLD_ENTER(zsb, obj_num);
err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db);
if (err) {
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
iput(ip);
return (err);
}
dmu_object_info_from_db(db, &doi);
if (doi.doi_bonus_type != DMU_OT_SA &&
(doi.doi_bonus_type != DMU_OT_ZNODE ||
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
iput(ip);
return (SET_ERROR(EINVAL));
}
hdl = dmu_buf_get_user(db);
if (hdl != NULL) {
if (ip == NULL) {
/*
* ilookup returned NULL, which means
* the znode is dying - but the SA handle isn't
* quite dead yet, we need to drop any locks
* we're holding, re-schedule the task and try again.
*/
sa_buf_rele(db, NULL);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
schedule();
goto again;
}
zp = sa_get_userdata(hdl);
/*
* Since "SA" does immediate eviction we
* should never find a sa handle that doesn't
* know about the znode.
*/
ASSERT3P(zp, !=, NULL);
mutex_enter(&zp->z_lock);
ASSERT3U(zp->z_id, ==, obj_num);
if (zp->z_unlinked) {
err = SET_ERROR(ENOENT);
} else {
igrab(ZTOI(zp));
*zpp = zp;
err = 0;
}
sa_buf_rele(db, NULL);
mutex_exit(&zp->z_lock);
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
iput(ip);
return (err);
}
ASSERT3P(ip, ==, NULL);
/*
* Not found create new znode/vnode but only if file exists.
*
* There is a small window where zfs_vget() could
* find this object while a file create is still in
* progress. This is checked for in zfs_znode_alloc()
*
* if zfs_znode_alloc() fails it will drop the hold on the
* bonus buffer.
*/
zp = zfs_znode_alloc(zsb, db, doi.doi_data_block_size,
doi.doi_bonus_type, obj_num, NULL, NULL);
if (zp == NULL) {
err = SET_ERROR(ENOENT);
} else {
*zpp = zp;
}
ZFS_OBJ_HOLD_EXIT(zsb, obj_num);
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;
rrw_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);
}
/*
* 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)
taskq_wait(dsl_pool_iput_taskq(dmu_objset_pool(zsb->z_os)));
/*
* 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);
rrw_exit(&zsb->z_teardown_lock, FTAG);
return (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.
*/
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) {
ASSERT(atomic_read(&ZTOI(zp)->i_count) > 0);
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;
rrw_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);
(void) dmu_objset_evict_dbufs(zsb->z_os);
return (0);
}