[mirror_zfs.git] / module / zfs / dmu_object.c

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2013 by Delphix. All rights reserved.
 */

#include <sys/dmu.h>
#include <sys/dmu_objset.h>
#include <sys/dmu_tx.h>
#include <sys/dnode.h>
#include <sys/zap.h>
#include <sys/zfeature.h>

uint64_t
dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize,
    dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
{
	uint64_t object;
	uint64_t L2_dnode_count = DNODES_PER_BLOCK <<
	    (DMU_META_DNODE(os)->dn_indblkshift - SPA_BLKPTRSHIFT);
	dnode_t *dn = NULL;
	int restarted = B_FALSE;

	mutex_enter(&os->os_obj_lock);
	for (;;) {
		object = os->os_obj_next;
		/*
		 * Each time we polish off an L2 bp worth of dnodes
		 * (2^13 objects), move to another L2 bp that's still
		 * reasonably sparse (at most 1/4 full).  Look from the
		 * beginning once, but after that keep looking from here.
		 * If we can't find one, just keep going from here.
		 */
		if (P2PHASE(object, L2_dnode_count) == 0) {
			uint64_t offset = restarted ? object << DNODE_SHIFT : 0;
			int error = dnode_next_offset(DMU_META_DNODE(os),
			    DNODE_FIND_HOLE,
			    &offset, 2, DNODES_PER_BLOCK >> 2, 0);
			restarted = B_TRUE;
			if (error == 0)
				object = offset >> DNODE_SHIFT;
		}
		os->os_obj_next = ++object;

		/*
		 * XXX We should check for an i/o error here and return
		 * up to our caller.  Actually we should pre-read it in
		 * dmu_tx_assign(), but there is currently no mechanism
		 * to do so.
		 */
		(void) dnode_hold_impl(os, object, DNODE_MUST_BE_FREE,
		    FTAG, &dn);
		if (dn)
			break;

		if (dmu_object_next(os, &object, B_TRUE, 0) == 0)
			os->os_obj_next = object - 1;
	}

	dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, tx);
	dnode_rele(dn, FTAG);

	mutex_exit(&os->os_obj_lock);

	dmu_tx_add_new_object(tx, os, object);
	return (object);
}

int
dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot,
    int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
{
	dnode_t *dn;
	int err;

	if (object == DMU_META_DNODE_OBJECT && !dmu_tx_private_ok(tx))
		return (SET_ERROR(EBADF));

	err = dnode_hold_impl(os, object, DNODE_MUST_BE_FREE, FTAG, &dn);
	if (err)
		return (err);
	dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, tx);
	dnode_rele(dn, FTAG);

	dmu_tx_add_new_object(tx, os, object);
	return (0);
}

int
dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot,
    int blocksize, dmu_object_type_t bonustype, int bonuslen)
{
	dnode_t *dn;
	dmu_tx_t *tx;
	int nblkptr;
	int err;

	if (object == DMU_META_DNODE_OBJECT)
		return (SET_ERROR(EBADF));

	err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED,
	    FTAG, &dn);
	if (err)
		return (err);

	if (dn->dn_type == ot && dn->dn_datablksz == blocksize &&
	    dn->dn_bonustype == bonustype && dn->dn_bonuslen == bonuslen) {
		/* nothing is changing, this is a noop */
		dnode_rele(dn, FTAG);
		return (0);
	}

	if (bonustype == DMU_OT_SA) {
		nblkptr = 1;
	} else {
		nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
	}

	/*
	 * If we are losing blkptrs or changing the block size this must
	 * be a new file instance.   We must clear out the previous file
	 * contents before we can change this type of metadata in the dnode.
	 */
	if (dn->dn_nblkptr > nblkptr || dn->dn_datablksz != blocksize) {
		err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END);
		if (err)
			goto out;
	}

	tx = dmu_tx_create(os);
	dmu_tx_hold_bonus(tx, object);
	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err) {
		dmu_tx_abort(tx);
		goto out;
	}

	dnode_reallocate(dn, ot, blocksize, bonustype, bonuslen, tx);

	dmu_tx_commit(tx);
out:
	dnode_rele(dn, FTAG);

	return (err);
}

int
dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx)
{
	dnode_t *dn;
	int err;

	ASSERT(object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx));

	err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED,
	    FTAG, &dn);
	if (err)
		return (err);

	ASSERT(dn->dn_type != DMU_OT_NONE);
	dnode_free_range(dn, 0, DMU_OBJECT_END, tx);
	dnode_free(dn, tx);
	dnode_rele(dn, FTAG);

	return (0);
}

int
dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole, uint64_t txg)
{
	uint64_t offset = (*objectp + 1) << DNODE_SHIFT;
	int error;

	error = dnode_next_offset(DMU_META_DNODE(os),
	    (hole ? DNODE_FIND_HOLE : 0), &offset, 0, DNODES_PER_BLOCK, txg);

	*objectp = offset >> DNODE_SHIFT;

	return (error);
}

/*
 * Turn this object from old_type into DMU_OTN_ZAP_METADATA, and bump the
 * refcount on SPA_FEATURE_EXTENSIBLE_DATASET.
 *
 * Only for use from syncing context, on MOS objects.
 */
void
dmu_object_zapify(objset_t *mos, uint64_t object, dmu_object_type_t old_type,
    dmu_tx_t *tx)
{
	dnode_t *dn;

	ASSERT(dmu_tx_is_syncing(tx));

	VERIFY0(dnode_hold(mos, object, FTAG, &dn));
	if (dn->dn_type == DMU_OTN_ZAP_METADATA) {
		dnode_rele(dn, FTAG);
		return;
	}
	ASSERT3U(dn->dn_type, ==, old_type);
	ASSERT0(dn->dn_maxblkid);
	dn->dn_next_type[tx->tx_txg & TXG_MASK] = dn->dn_type =
	    DMU_OTN_ZAP_METADATA;
	dnode_setdirty(dn, tx);
	dnode_rele(dn, FTAG);

	mzap_create_impl(mos, object, 0, 0, tx);

	spa_feature_incr(dmu_objset_spa(mos),
	    SPA_FEATURE_EXTENSIBLE_DATASET, tx);
}

void
dmu_object_free_zapified(objset_t *mos, uint64_t object, dmu_tx_t *tx)
{
	dnode_t *dn;
	dmu_object_type_t t;

	ASSERT(dmu_tx_is_syncing(tx));

	VERIFY0(dnode_hold(mos, object, FTAG, &dn));
	t = dn->dn_type;
	dnode_rele(dn, FTAG);

	if (t == DMU_OTN_ZAP_METADATA) {
		spa_feature_decr(dmu_objset_spa(mos),
		    SPA_FEATURE_EXTENSIBLE_DATASET, tx);
	}
	VERIFY0(dmu_object_free(mos, object, tx));
}

#if defined(_KERNEL) && defined(HAVE_SPL)
EXPORT_SYMBOL(dmu_object_alloc);
EXPORT_SYMBOL(dmu_object_claim);
EXPORT_SYMBOL(dmu_object_reclaim);
EXPORT_SYMBOL(dmu_object_free);
EXPORT_SYMBOL(dmu_object_next);
EXPORT_SYMBOL(dmu_object_zapify);
EXPORT_SYMBOL(dmu_object_free_zapified);
#endif
Commit	Line	Data
34dc7c2f BB	1	/*
	2	* CDDL HEADER START
	3	*
	4	* The contents of this file are subject to the terms of the
	5	* Common Development and Distribution License (the "License").
	6	* You may not use this file except in compliance with the License.
	7	*
	8	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	9	* or http://www.opensolaris.org/os/licensing.
	10	* See the License for the specific language governing permissions
	11	* and limitations under the License.
	12	*
	13	* When distributing Covered Code, include this CDDL HEADER in each
	14	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	15	* If applicable, add the following below this CDDL HEADER, with the
	16	* fields enclosed by brackets "[]" replaced with your own identifying
	17	* information: Portions Copyright [yyyy] [name of copyright owner]
	18	*
	19	* CDDL HEADER END
	20	*/
	21	/*
428870ff	22	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
2e528b49	23	* Copyright (c) 2013 by Delphix. All rights reserved.
34dc7c2f BB	24	*/
34dc7c2f BB	25
34dc7c2f BB	26	#include <sys/dmu.h>
	27	#include <sys/dmu_objset.h>
	28	#include <sys/dmu_tx.h>
	29	#include <sys/dnode.h>
fa86b5db MA	30	#include <sys/zap.h>
fa86b5db MA	31	#include <sys/zfeature.h>
34dc7c2f BB	32
	33	uint64_t
	34	dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize,
	35	dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
	36	{
34dc7c2f BB	37	uint64_t object;
34dc7c2f BB	38	uint64_t L2_dnode_count = DNODES_PER_BLOCK <<
572e2857	39	(DMU_META_DNODE(os)->dn_indblkshift - SPA_BLKPTRSHIFT);
34dc7c2f BB	40	dnode_t *dn = NULL;
	41	int restarted = B_FALSE;
	42
428870ff	43	mutex_enter(&os->os_obj_lock);
34dc7c2f	44	for (;;) {
428870ff	45	object = os->os_obj_next;
34dc7c2f BB	46	/*
	47	* Each time we polish off an L2 bp worth of dnodes
	48	* (2^13 objects), move to another L2 bp that's still
	49	* reasonably sparse (at most 1/4 full). Look from the
	50	* beginning once, but after that keep looking from here.
	51	* If we can't find one, just keep going from here.
	52	*/
	53	if (P2PHASE(object, L2_dnode_count) == 0) {
	54	uint64_t offset = restarted ? object << DNODE_SHIFT : 0;
572e2857	55	int error = dnode_next_offset(DMU_META_DNODE(os),
b128c09f BB	56	DNODE_FIND_HOLE,
b128c09f BB	57	&offset, 2, DNODES_PER_BLOCK >> 2, 0);
34dc7c2f BB	58	restarted = B_TRUE;
	59	if (error == 0)
	60	object = offset >> DNODE_SHIFT;
	61	}
428870ff	62	os->os_obj_next = ++object;
34dc7c2f BB	63
	64	/*
	65	* XXX We should check for an i/o error here and return
	66	* up to our caller. Actually we should pre-read it in
	67	* dmu_tx_assign(), but there is currently no mechanism
	68	* to do so.
	69	*/
428870ff	70	(void) dnode_hold_impl(os, object, DNODE_MUST_BE_FREE,
34dc7c2f BB	71	FTAG, &dn);
	72	if (dn)
	73	break;
	74
	75	if (dmu_object_next(os, &object, B_TRUE, 0) == 0)
428870ff	76	os->os_obj_next = object - 1;
34dc7c2f BB	77	}
	78
	79	dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, tx);
	80	dnode_rele(dn, FTAG);
	81
428870ff	82	mutex_exit(&os->os_obj_lock);
34dc7c2f BB	83
	84	dmu_tx_add_new_object(tx, os, object);
	85	return (object);
	86	}
	87
	88	int
	89	dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot,
	90	int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx)
	91	{
	92	dnode_t *dn;
	93	int err;
	94
	95	if (object == DMU_META_DNODE_OBJECT && !dmu_tx_private_ok(tx))
2e528b49	96	return (SET_ERROR(EBADF));
34dc7c2f	97
428870ff	98	err = dnode_hold_impl(os, object, DNODE_MUST_BE_FREE, FTAG, &dn);
34dc7c2f BB	99	if (err)
	100	return (err);
	101	dnode_allocate(dn, ot, blocksize, 0, bonustype, bonuslen, tx);
	102	dnode_rele(dn, FTAG);
	103
	104	dmu_tx_add_new_object(tx, os, object);
	105	return (0);
	106	}
	107
	108	int
	109	dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot,
9babb374	110	int blocksize, dmu_object_type_t bonustype, int bonuslen)
34dc7c2f BB	111	{
34dc7c2f BB	112	dnode_t *dn;
9babb374 BB	113	dmu_tx_t *tx;
9babb374 BB	114	int nblkptr;
34dc7c2f BB	115	int err;
34dc7c2f BB	116
9babb374	117	if (object == DMU_META_DNODE_OBJECT)
2e528b49	118	return (SET_ERROR(EBADF));
34dc7c2f	119
428870ff	120	err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED,
34dc7c2f BB	121	FTAG, &dn);
	122	if (err)
	123	return (err);
9babb374 BB	124
	125	if (dn->dn_type == ot && dn->dn_datablksz == blocksize &&
	126	dn->dn_bonustype == bonustype && dn->dn_bonuslen == bonuslen) {
	127	/* nothing is changing, this is a noop */
	128	dnode_rele(dn, FTAG);
	129	return (0);
	130	}
	131
428870ff BB	132	if (bonustype == DMU_OT_SA) {
	133	nblkptr = 1;
	134	} else {
	135	nblkptr = 1 + ((DN_MAX_BONUSLEN - bonuslen) >> SPA_BLKPTRSHIFT);
	136	}
9babb374 BB	137
	138	/*
	139	* If we are losing blkptrs or changing the block size this must
	140	* be a new file instance. We must clear out the previous file
	141	* contents before we can change this type of metadata in the dnode.
	142	*/
	143	if (dn->dn_nblkptr > nblkptr \|\| dn->dn_datablksz != blocksize) {
	144	err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END);
	145	if (err)
	146	goto out;
	147	}
	148
	149	tx = dmu_tx_create(os);
	150	dmu_tx_hold_bonus(tx, object);
	151	err = dmu_tx_assign(tx, TXG_WAIT);
	152	if (err) {
	153	dmu_tx_abort(tx);
	154	goto out;
	155	}
	156
34dc7c2f	157	dnode_reallocate(dn, ot, blocksize, bonustype, bonuslen, tx);
9babb374 BB	158
	159	dmu_tx_commit(tx);
	160	out:
34dc7c2f BB	161	dnode_rele(dn, FTAG);
34dc7c2f BB	162
9babb374	163	return (err);
34dc7c2f BB	164	}
	165
	166	int
	167	dmu_object_free(objset_t os, uint64_t object, dmu_tx_t tx)
	168	{
	169	dnode_t *dn;
	170	int err;
	171
	172	ASSERT(object != DMU_META_DNODE_OBJECT \|\| dmu_tx_private_ok(tx));
	173
428870ff	174	err = dnode_hold_impl(os, object, DNODE_MUST_BE_ALLOCATED,
34dc7c2f BB	175	FTAG, &dn);
	176	if (err)
	177	return (err);
	178
	179	ASSERT(dn->dn_type != DMU_OT_NONE);
b128c09f	180	dnode_free_range(dn, 0, DMU_OBJECT_END, tx);
34dc7c2f BB	181	dnode_free(dn, tx);
	182	dnode_rele(dn, FTAG);
	183
	184	return (0);
	185	}
	186
	187	int
	188	dmu_object_next(objset_t os, uint64_t objectp, boolean_t hole, uint64_t txg)
	189	{
	190	uint64_t offset = (*objectp + 1) << DNODE_SHIFT;
	191	int error;
	192
572e2857	193	error = dnode_next_offset(DMU_META_DNODE(os),
b128c09f	194	(hole ? DNODE_FIND_HOLE : 0), &offset, 0, DNODES_PER_BLOCK, txg);
34dc7c2f BB	195
	196	*objectp = offset >> DNODE_SHIFT;
	197
	198	return (error);
	199	}
c28b2279	200
fa86b5db MA	201	/*
	202	* Turn this object from old_type into DMU_OTN_ZAP_METADATA, and bump the
	203	* refcount on SPA_FEATURE_EXTENSIBLE_DATASET.
	204	*
	205	* Only for use from syncing context, on MOS objects.
	206	*/
	207	void
	208	dmu_object_zapify(objset_t *mos, uint64_t object, dmu_object_type_t old_type,
	209	dmu_tx_t *tx)
	210	{
	211	dnode_t *dn;
	212
	213	ASSERT(dmu_tx_is_syncing(tx));
	214
	215	VERIFY0(dnode_hold(mos, object, FTAG, &dn));
	216	if (dn->dn_type == DMU_OTN_ZAP_METADATA) {
	217	dnode_rele(dn, FTAG);
	218	return;
	219	}
	220	ASSERT3U(dn->dn_type, ==, old_type);
	221	ASSERT0(dn->dn_maxblkid);
	222	dn->dn_next_type[tx->tx_txg & TXG_MASK] = dn->dn_type =
	223	DMU_OTN_ZAP_METADATA;
	224	dnode_setdirty(dn, tx);
	225	dnode_rele(dn, FTAG);
	226
	227	mzap_create_impl(mos, object, 0, 0, tx);
	228
	229	spa_feature_incr(dmu_objset_spa(mos),
	230	SPA_FEATURE_EXTENSIBLE_DATASET, tx);
	231	}
	232
	233	void
	234	dmu_object_free_zapified(objset_t mos, uint64_t object, dmu_tx_t tx)
	235	{
	236	dnode_t *dn;
	237	dmu_object_type_t t;
	238
	239	ASSERT(dmu_tx_is_syncing(tx));
	240
	241	VERIFY0(dnode_hold(mos, object, FTAG, &dn));
	242	t = dn->dn_type;
	243	dnode_rele(dn, FTAG);
	244
	245	if (t == DMU_OTN_ZAP_METADATA) {
	246	spa_feature_decr(dmu_objset_spa(mos),
	247	SPA_FEATURE_EXTENSIBLE_DATASET, tx);
	248	}
	249	VERIFY0(dmu_object_free(mos, object, tx));
	250	}
	251
c28b2279 BB	252	#if defined(_KERNEL) && defined(HAVE_SPL)
	253	EXPORT_SYMBOL(dmu_object_alloc);
	254	EXPORT_SYMBOL(dmu_object_claim);
	255	EXPORT_SYMBOL(dmu_object_reclaim);
	256	EXPORT_SYMBOL(dmu_object_free);
	257	EXPORT_SYMBOL(dmu_object_next);
fa86b5db MA	258	EXPORT_SYMBOL(dmu_object_zapify);
fa86b5db MA	259	EXPORT_SYMBOL(dmu_object_free_zapified);
c28b2279	260	#endif