int
dmu_objset_destroy(const char *name)
{
objset_t *os;
int error;
/*
* If it looks like we'll be able to destroy it, and there's
* an unplayed replay log sitting around, destroy the log.
* It would be nicer to do this in dsl_dataset_destroy_sync(),
* but the replay log objset is modified in open context.
*/
error = dmu_objset_open(name, DMU_OST_ANY,
DS_MODE_OWNER|DS_MODE_READONLY|DS_MODE_INCONSISTENT, &os);
if (error == 0) {
dsl_dataset_t *ds = os->os->os_dsl_dataset;
zil_destroy(dmu_objset_zil(os), B_FALSE);
error = dsl_dataset_destroy(ds, os);
/*
* dsl_dataset_destroy() closes the ds.
*/
kmem_free(os, sizeof (objset_t));
}
return (error);
}
/* ARGSUSED */
int
zil_clear_log_chain(char *osname, void *txarg)
{
zilog_t *zilog;
zil_header_t *zh;
objset_t *os;
dmu_tx_t *tx;
int error;
error = dmu_objset_open(osname, DMU_OST_ANY, DS_MODE_USER, &os);
if (error) {
cmn_err(CE_WARN, "can't open objset for %s", osname);
return (0);
}
zilog = dmu_objset_zil(os);
tx = dmu_tx_create(zilog->zl_os);
(void) dmu_tx_assign(tx, TXG_WAIT);
zh = zil_header_in_syncing_context(zilog);
BP_ZERO(&zh->zh_log);
dsl_dataset_dirty(dmu_objset_ds(os), tx);
dmu_tx_commit(tx);
dmu_objset_close(os);
return (0);
}
/*
* Open an intent log.
*/
zilog_t *
zil_open(objset_t *os, zil_get_data_t *get_data)
{
zilog_t *zilog = dmu_objset_zil(os);
zilog->zl_get_data = get_data;
zilog->zl_clean_taskq = taskq_create("zil_clean", 1, minclsyspri,
2, 2, TASKQ_PREPOPULATE);
return (zilog);
}
int
dmu_objset_snapshot(char *fsname, char *snapname, boolean_t recursive)
{
dsl_sync_task_t *dst;
struct osnode *osn;
struct snaparg sn = { 0 };
spa_t *spa;
int err;
(void) strcpy(sn.failed, fsname);
err = spa_open(fsname, &spa, FTAG);
if (err)
return (err);
sn.dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
sn.snapname = snapname;
list_create(&sn.objsets, sizeof (struct osnode),
offsetof(struct osnode, node));
if (recursive) {
sn.checkperms = B_TRUE;
err = dmu_objset_find(fsname,
dmu_objset_snapshot_one, &sn, DS_FIND_CHILDREN);
} else {
sn.checkperms = B_FALSE;
err = dmu_objset_snapshot_one(fsname, &sn);
}
if (err)
goto out;
err = dsl_sync_task_group_wait(sn.dstg);
for (dst = list_head(&sn.dstg->dstg_tasks); dst;
dst = list_next(&sn.dstg->dstg_tasks, dst)) {
dsl_dataset_t *ds = dst->dst_arg1;
if (dst->dst_err)
dsl_dataset_name(ds, sn.failed);
}
out:
while (osn = list_head(&sn.objsets)) {
list_remove(&sn.objsets, osn);
zil_resume(dmu_objset_zil(osn->os));
dmu_objset_close(osn->os);
kmem_free(osn, sizeof (struct osnode));
}
list_destroy(&sn.objsets);
if (err)
(void) strcpy(fsname, sn.failed);
dsl_sync_task_group_destroy(sn.dstg);
spa_close(spa, FTAG);
return (err);
}
int
zil_claim(char *osname, void *txarg)
{
dmu_tx_t *tx = txarg;
uint64_t first_txg = dmu_tx_get_txg(tx);
zilog_t *zilog;
zil_header_t *zh;
objset_t *os;
int error;
error = dmu_objset_open(osname, DMU_OST_ANY, DS_MODE_USER, &os);
if (error) {
cmn_err(CE_WARN, "can't open objset for %s", osname);
return (0);
}
zilog = dmu_objset_zil(os);
zh = zil_header_in_syncing_context(zilog);
/*
* Record here whether the zil has any records to replay.
* If the header block pointer is null or the block points
* to the stubby then we know there are no valid log records.
* We use the header to store this state as the the zilog gets
* freed later in dmu_objset_close().
* The flags (and the rest of the header fields) are cleared in
* zil_sync() as a result of a zil_destroy(), after replaying the log.
*
* Note, the intent log can be empty but still need the
* stubby to be claimed.
*/
if (!zil_empty(zilog))
zh->zh_flags |= ZIL_REPLAY_NEEDED;
/*
* Claim all log blocks if we haven't already done so, and remember
* the highest claimed sequence number. This ensures that if we can
* read only part of the log now (e.g. due to a missing device),
* but we can read the entire log later, we will not try to replay
* or destroy beyond the last block we successfully claimed.
*/
ASSERT3U(zh->zh_claim_txg, <=, first_txg);
if (zh->zh_claim_txg == 0 && !BP_IS_HOLE(&zh->zh_log)) {
zh->zh_claim_txg = first_txg;
zh->zh_claim_seq = zil_parse(zilog, zil_claim_log_block,
zil_claim_log_record, tx, first_txg);
dsl_dataset_dirty(dmu_objset_ds(os), tx);
}
ASSERT3U(first_txg, ==, (spa_last_synced_txg(zilog->zl_spa) + 1));
dmu_objset_close(os);
return (0);
}
int
dmu_objset_snapshot(char *fsname, char *snapname,
nvlist_t *props, boolean_t recursive)
{
dsl_sync_task_t *dst;
struct snaparg *sn;
spa_t *spa;
int err;
sn = kmem_alloc(sizeof (struct snaparg), KM_SLEEP);
(void) strcpy(sn->failed, fsname);
err = spa_open(fsname, &spa, FTAG);
if (err) {
kmem_free(sn, sizeof (struct snaparg));
return (err);
}
sn->dstg = dsl_sync_task_group_create(spa_get_dsl(spa));
sn->snapname = snapname;
sn->props = props;
if (recursive) {
sn->checkperms = B_TRUE;
err = dmu_objset_find(fsname,
dmu_objset_snapshot_one, sn, DS_FIND_CHILDREN);
} else {
sn->checkperms = B_FALSE;
err = dmu_objset_snapshot_one(fsname, sn);
}
if (err == 0)
err = dsl_sync_task_group_wait(sn->dstg);
for (dst = list_head(&sn->dstg->dstg_tasks); dst;
dst = list_next(&sn->dstg->dstg_tasks, dst)) {
objset_t *os = dst->dst_arg1;
dsl_dataset_t *ds = os->os->os_dsl_dataset;
if (dst->dst_err)
dsl_dataset_name(ds, sn->failed);
zil_resume(dmu_objset_zil(os));
dmu_objset_close(os);
}
if (err)
(void) strcpy(fsname, sn->failed);
dsl_sync_task_group_destroy(sn->dstg);
spa_close(spa, FTAG);
kmem_free(sn, sizeof (struct snaparg));
return (err);
}
static int
dmu_objset_snapshot_one(char *name, void *arg)
{
struct snaparg *sn = arg;
objset_t *os;
int err;
(void) strcpy(sn->failed, name);
/*
* Check permissions only when requested. This only applies when
* doing a recursive snapshot. The permission checks for the starting
* dataset have already been performed in zfs_secpolicy_snapshot()
*/
if (sn->checkperms == B_TRUE &&
(err = zfs_secpolicy_snapshot_perms(name, CRED())))
return (err);
err = dmu_objset_open(name, DMU_OST_ANY, DS_MODE_USER, &os);
if (err != 0)
return (err);
/* If the objset is in an inconsistent state, return busy */
if (os->os->os_dsl_dataset->ds_phys->ds_flags & DS_FLAG_INCONSISTENT) {
dmu_objset_close(os);
return (EBUSY);
}
/*
* NB: we need to wait for all in-flight changes to get to disk,
* so that we snapshot those changes. zil_suspend does this as
* a side effect.
*/
err = zil_suspend(dmu_objset_zil(os));
if (err == 0) {
struct osnode *osn;
dsl_sync_task_create(sn->dstg, dsl_dataset_snapshot_check,
dsl_dataset_snapshot_sync, os->os->os_dsl_dataset,
sn->snapname, 3);
osn = kmem_alloc(sizeof (struct osnode), KM_SLEEP);
osn->os = os;
list_insert_tail(&sn->objsets, osn);
} else {
dmu_objset_close(os);
}
return (err);
}
/* ARGSUSED */
int
zil_vdev_offline(char *osname, void *arg)
{
objset_t *os;
zilog_t *zilog;
int error;
error = dmu_objset_open(osname, DMU_OST_ANY, DS_MODE_USER, &os);
if (error)
return (error);
zilog = dmu_objset_zil(os);
if (zil_suspend(zilog) != 0)
error = EEXIST;
else
zil_resume(zilog);
dmu_objset_close(os);
return (error);
}
/* ARGSUSED */
int
zil_check_log_chain(char *osname, void *txarg)
{
zilog_t *zilog;
zil_header_t *zh;
blkptr_t blk;
arc_buf_t *abuf;
objset_t *os;
char *lrbuf;
zil_trailer_t *ztp;
int error;
error = dmu_objset_open(osname, DMU_OST_ANY, DS_MODE_USER, &os);
if (error) {
cmn_err(CE_WARN, "can't open objset for %s", osname);
return (0);
}
zilog = dmu_objset_zil(os);
zh = zil_header_in_syncing_context(zilog);
blk = zh->zh_log;
if (BP_IS_HOLE(&blk)) {
dmu_objset_close(os);
return (0); /* no chain */
}
for (;;) {
error = zil_read_log_block(zilog, &blk, &abuf);
if (error)
break;
lrbuf = abuf->b_data;
ztp = (zil_trailer_t *)(lrbuf + BP_GET_LSIZE(&blk)) - 1;
blk = ztp->zit_next_blk;
VERIFY(arc_buf_remove_ref(abuf, &abuf) == 1);
}
dmu_objset_close(os);
if (error == ECKSUM)
return (0); /* normal end of chain */
return (error);
}
/*
* If this dataset has a non-empty intent log, replay it and destroy it.
*/
void
zil_replay(objset_t *os, void *arg, zil_replay_func_t *replay_func[TX_MAX_TYPE])
{
zilog_t *zilog = dmu_objset_zil(os);
const zil_header_t *zh = zilog->zl_header;
zil_replay_arg_t zr;
if ((zh->zh_flags & ZIL_REPLAY_NEEDED) == 0) {
zil_destroy(zilog, B_TRUE);
return;
}
zr.zr_os = os;
zr.zr_replay = replay_func;
zr.zr_arg = arg;
zr.zr_byteswap = BP_SHOULD_BYTESWAP(&zh->zh_log);
/* XXX: Changed to use vmem_alloc instead of kmem_alloc for
* large allocation size (I think this is safe here).
*/
zr.zr_lrbuf = vmem_alloc(2 * SPA_MAXBLOCKSIZE, KM_SLEEP);
/*
* Wait for in-progress removes to sync before starting replay.
*/
txg_wait_synced(zilog->zl_dmu_pool, 0);
zilog->zl_replay = B_TRUE;
zilog->zl_replay_time = lbolt;
ASSERT(zilog->zl_replay_blks == 0);
(void) zil_parse(zilog, zil_incr_blks, zil_replay_log_record, &zr,
zh->zh_claim_txg);
vmem_free(zr.zr_lrbuf, 2 * SPA_MAXBLOCKSIZE);
zil_destroy(zilog, B_FALSE);
txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg);
zilog->zl_replay = B_FALSE;
}
/*
* If this dataset has a non-empty intent log, replay it and destroy it.
*/
void
zil_replay(objset_t *os, void *arg, zil_replay_func_t *replay_func[TX_MAX_TYPE])
{
zilog_t *zilog = dmu_objset_zil(os);
const zil_header_t *zh = zilog->zl_header;
zil_replay_arg_t zr;
if ((zh->zh_flags & ZIL_REPLAY_NEEDED) == 0) {
zil_destroy(zilog, B_TRUE);
return;
}
//printf("ZFS: Replaying ZIL on %s...\n", os->os->os_spa->spa_name);
zr.zr_os = os;
zr.zr_replay = replay_func;
zr.zr_arg = arg;
zr.zr_byteswap = BP_SHOULD_BYTESWAP(&zh->zh_log);
zr.zr_lrbuf = kmem_alloc(2 * SPA_MAXBLOCKSIZE, KM_SLEEP);
/*
* Wait for in-progress removes to sync before starting replay.
*/
txg_wait_synced(zilog->zl_dmu_pool, 0);
zilog->zl_replay = B_TRUE;
zilog->zl_replay_time = LBOLT;
ASSERT(zilog->zl_replay_blks == 0);
(void) zil_parse(zilog, zil_incr_blks, zil_replay_log_record, &zr,
zh->zh_claim_txg);
kmem_free(zr.zr_lrbuf, 2 * SPA_MAXBLOCKSIZE);
zil_destroy(zilog, B_FALSE);
txg_wait_synced(zilog->zl_dmu_pool, zilog->zl_destroy_txg);
zilog->zl_replay = B_FALSE;
//printf("ZFS: Replay of ZIL on %s finished.\n", os->os->os_spa->spa_name);
}
/*
* Create a block device minor node and setup the linkage between it
* and the specified volume. Once this function returns the block
* device is live and ready for use.
*/
static int
zvol_create_minor_impl(const char *name)
{
zvol_state_t *zv;
objset_t *os;
dmu_object_info_t *doi;
uint64_t volsize;
uint64_t len;
unsigned minor = 0;
int error = 0;
mutex_enter(&zvol_state_lock);
zv = zvol_find_by_name(name);
if (zv) {
error = SET_ERROR(EEXIST);
goto out;
}
doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP);
error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, zvol_tag, &os);
if (error)
goto out_doi;
error = dmu_object_info(os, ZVOL_OBJ, doi);
if (error)
goto out_dmu_objset_disown;
error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
if (error)
goto out_dmu_objset_disown;
error = zvol_find_minor(&minor);
if (error)
goto out_dmu_objset_disown;
zv = zvol_alloc(MKDEV(zvol_major, minor), name);
if (zv == NULL) {
error = SET_ERROR(EAGAIN);
goto out_dmu_objset_disown;
}
if (dmu_objset_is_snapshot(os))
zv->zv_flags |= ZVOL_RDONLY;
zv->zv_volblocksize = doi->doi_data_block_size;
zv->zv_volsize = volsize;
zv->zv_objset = os;
set_capacity(zv->zv_disk, zv->zv_volsize >> 9);
blk_queue_max_hw_sectors(zv->zv_queue, (DMU_MAX_ACCESS / 4) >> 9);
blk_queue_max_segments(zv->zv_queue, UINT16_MAX);
blk_queue_max_segment_size(zv->zv_queue, UINT_MAX);
blk_queue_physical_block_size(zv->zv_queue, zv->zv_volblocksize);
blk_queue_io_opt(zv->zv_queue, zv->zv_volblocksize);
blk_queue_max_discard_sectors(zv->zv_queue,
(zvol_max_discard_blocks * zv->zv_volblocksize) >> 9);
blk_queue_discard_granularity(zv->zv_queue, zv->zv_volblocksize);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zv->zv_queue);
#ifdef QUEUE_FLAG_NONROT
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zv->zv_queue);
#endif
#ifdef QUEUE_FLAG_ADD_RANDOM
queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, zv->zv_queue);
#endif
if (spa_writeable(dmu_objset_spa(os))) {
if (zil_replay_disable)
zil_destroy(dmu_objset_zil(os), B_FALSE);
else
zil_replay(os, zv, zvol_replay_vector);
}
/*
* When udev detects the addition of the device it will immediately
* invoke blkid(8) to determine the type of content on the device.
* Prefetching the blocks commonly scanned by blkid(8) will speed
* up this process.
*/
len = MIN(MAX(zvol_prefetch_bytes, 0), SPA_MAXBLOCKSIZE);
if (len > 0) {
dmu_prefetch(os, ZVOL_OBJ, 0, 0, len, ZIO_PRIORITY_SYNC_READ);
dmu_prefetch(os, ZVOL_OBJ, 0, volsize - len, len,
ZIO_PRIORITY_SYNC_READ);
}
zv->zv_objset = NULL;
out_dmu_objset_disown:
dmu_objset_disown(os, zvol_tag);
out_doi:
kmem_free(doi, sizeof (dmu_object_info_t));
out:
if (error == 0) {
zvol_insert(zv);
/*
* Drop the lock to prevent deadlock with sys_open() ->
* zvol_open(), which first takes bd_disk->bd_mutex and then
* takes zvol_state_lock, whereas this code path first takes
* zvol_state_lock, and then takes bd_disk->bd_mutex.
*/
mutex_exit(&zvol_state_lock);
add_disk(zv->zv_disk);
} else {
mutex_exit(&zvol_state_lock);
}
return (SET_ERROR(error));
}
static int
__zvol_create_minor(const char *name)
{
zvol_state_t *zv;
objset_t *os;
dmu_object_info_t *doi;
uint64_t volsize;
unsigned minor = 0;
int error = 0;
ASSERT(MUTEX_HELD(&zvol_state_lock));
zv = zvol_find_by_name(name);
if (zv) {
error = EEXIST;
goto out;
}
doi = kmem_alloc(sizeof(dmu_object_info_t), KM_SLEEP);
error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, zvol_tag, &os);
if (error)
goto out_doi;
/* Make sure we have the key loaded if we need one. */
error = dsl_crypto_key_inherit(name);
if (error != 0 && error != EEXIST)
goto out_dmu_objset_disown;
error = dmu_object_info(os, ZVOL_OBJ, doi);
if (error)
goto out_dmu_objset_disown;
error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
if (error)
goto out_dmu_objset_disown;
error = zvol_find_minor(&minor);
if (error)
goto out_dmu_objset_disown;
zv = zvol_alloc(MKDEV(zvol_major, minor), name);
if (zv == NULL) {
error = EAGAIN;
goto out_dmu_objset_disown;
}
if (dmu_objset_is_snapshot(os))
zv->zv_flags |= ZVOL_RDONLY;
zv->zv_volblocksize = doi->doi_data_block_size;
zv->zv_volsize = volsize;
zv->zv_objset = os;
set_capacity(zv->zv_disk, zv->zv_volsize >> 9);
blk_queue_max_hw_sectors(zv->zv_queue, UINT_MAX);
blk_queue_max_segments(zv->zv_queue, UINT16_MAX);
blk_queue_max_segment_size(zv->zv_queue, UINT_MAX);
blk_queue_physical_block_size(zv->zv_queue, zv->zv_volblocksize);
blk_queue_io_opt(zv->zv_queue, zv->zv_volblocksize);
#ifdef HAVE_BLK_QUEUE_DISCARD
blk_queue_max_discard_sectors(zv->zv_queue,
(zvol_max_discard_blocks * zv->zv_volblocksize) >> 9);
blk_queue_discard_granularity(zv->zv_queue, zv->zv_volblocksize);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zv->zv_queue);
#endif
#ifdef HAVE_BLK_QUEUE_NONROT
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zv->zv_queue);
#endif
if (zil_replay_disable)
zil_destroy(dmu_objset_zil(os), B_FALSE);
else
zil_replay(os, zv, zvol_replay_vector);
out_dmu_objset_disown:
dmu_objset_disown(os, zvol_tag);
zv->zv_objset = NULL;
out_doi:
kmem_free(doi, sizeof(dmu_object_info_t));
out:
if (error == 0) {
zvol_insert(zv);
add_disk(zv->zv_disk);
}
return (error);
}
static int
__zvol_create_minor(const char *name)
{
zvol_state_t *zv;
objset_t *os;
dmu_object_info_t *doi;
uint64_t volsize;
unsigned minor = 0;
int error = 0;
ASSERT(MUTEX_HELD(&zvol_state_lock));
zv = zvol_find_by_name(name);
if (zv) {
error = EEXIST;
goto out;
}
doi = kmem_alloc(sizeof(dmu_object_info_t), KM_SLEEP);
error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, zvol_tag, &os);
if (error)
goto out_doi;
error = dmu_object_info(os, ZVOL_OBJ, doi);
if (error)
goto out_dmu_objset_disown;
error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
if (error)
goto out_dmu_objset_disown;
error = zvol_find_minor(&minor);
if (error)
goto out_dmu_objset_disown;
zv = zvol_alloc(MKDEV(zvol_major, minor), name);
if (zv == NULL) {
error = EAGAIN;
goto out_dmu_objset_disown;
}
if (dmu_objset_is_snapshot(os))
zv->zv_flags |= ZVOL_RDONLY;
zv->zv_volblocksize = doi->doi_data_block_size;
zv->zv_volsize = volsize;
zv->zv_objset = os;
set_capacity(zv->zv_disk, zv->zv_volsize >> 9);
if (zil_replay_disable)
zil_destroy(dmu_objset_zil(os), B_FALSE);
else
zil_replay(os, zv, zvol_replay_vector);
out_dmu_objset_disown:
dmu_objset_disown(os, zvol_tag);
zv->zv_objset = NULL;
out_doi:
kmem_free(doi, sizeof(dmu_object_info_t));
out:
if (error == 0) {
zvol_insert(zv);
add_disk(zv->zv_disk);
}
return (error);
}
