static void
trim_thread(void *arg)
{
spa_t *spa = arg;
zio_t *zio;
for (;;) {
mutex_enter(&spa->spa_trim_lock);
if (spa->spa_trim_thread == NULL) {
spa->spa_trim_thread = curthread;
cv_signal(&spa->spa_trim_cv);
mutex_exit(&spa->spa_trim_lock);
thread_exit();
}
cv_wait(&spa->spa_trim_cv, &spa->spa_trim_lock);
mutex_exit(&spa->spa_trim_lock);
zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
trim_map_commit(spa, zio, spa->spa_root_vdev);
(void) zio_wait(zio);
trim_map_commit_done(spa, spa->spa_root_vdev);
spa_config_exit(spa, SCL_STATE, FTAG);
}
}
static void
trim_thread(void *arg)
{
spa_t *spa = arg;
zio_t *zio;
#ifdef _KERNEL
(void) snprintf(curthread->td_name, sizeof(curthread->td_name),
"trim %s", spa_name(spa));
#endif
for (;;) {
mutex_enter(&spa->spa_trim_lock);
if (spa->spa_trim_thread == NULL) {
spa->spa_trim_thread = curthread;
cv_signal(&spa->spa_trim_cv);
mutex_exit(&spa->spa_trim_lock);
thread_exit();
}
(void) cv_timedwait(&spa->spa_trim_cv, &spa->spa_trim_lock,
hz * trim_max_interval);
mutex_exit(&spa->spa_trim_lock);
zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
trim_map_commit(spa, zio, spa->spa_root_vdev);
(void) zio_wait(zio);
trim_map_commit_done(spa, spa->spa_root_vdev);
spa_config_exit(spa, SCL_STATE, FTAG);
}
}
/* Takes care of physical writing and limiting # of concurrent ZIOs. */
static int
vdev_initialize_write(vdev_t *vd, uint64_t start, uint64_t size, abd_t *data)
{
spa_t *spa = vd->vdev_spa;
/* Limit inflight initializing I/Os */
mutex_enter(&vd->vdev_initialize_io_lock);
while (vd->vdev_initialize_inflight >= zfs_initialize_limit) {
cv_wait(&vd->vdev_initialize_io_cv,
&vd->vdev_initialize_io_lock);
}
vd->vdev_initialize_inflight++;
mutex_exit(&vd->vdev_initialize_io_lock);
dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
VERIFY0(dmu_tx_assign(tx, TXG_WAIT));
uint64_t txg = dmu_tx_get_txg(tx);
spa_config_enter(spa, SCL_STATE_ALL, vd, RW_READER);
mutex_enter(&vd->vdev_initialize_lock);
if (vd->vdev_initialize_offset[txg & TXG_MASK] == 0) {
uint64_t *guid = kmem_zalloc(sizeof (uint64_t), KM_SLEEP);
*guid = vd->vdev_guid;
/* This is the first write of this txg. */
dsl_sync_task_nowait(spa_get_dsl(spa),
vdev_initialize_zap_update_sync, guid, 2,
ZFS_SPACE_CHECK_RESERVED, tx);
}
/*
* We know the vdev struct will still be around since all
* consumers of vdev_free must stop the initialization first.
*/
if (vdev_initialize_should_stop(vd)) {
mutex_enter(&vd->vdev_initialize_io_lock);
ASSERT3U(vd->vdev_initialize_inflight, >, 0);
vd->vdev_initialize_inflight--;
mutex_exit(&vd->vdev_initialize_io_lock);
spa_config_exit(vd->vdev_spa, SCL_STATE_ALL, vd);
mutex_exit(&vd->vdev_initialize_lock);
dmu_tx_commit(tx);
return (SET_ERROR(EINTR));
}
mutex_exit(&vd->vdev_initialize_lock);
vd->vdev_initialize_offset[txg & TXG_MASK] = start + size;
zio_nowait(zio_write_phys(spa->spa_txg_zio[txg & TXG_MASK], vd, start,
size, data, ZIO_CHECKSUM_OFF, vdev_initialize_cb, NULL,
ZIO_PRIORITY_INITIALIZING, ZIO_FLAG_CANFAIL, B_FALSE));
/* vdev_initialize_cb releases SCL_STATE_ALL */
dmu_tx_commit(tx);
return (0);
}
/*
* Choose a random vdev, label, and MMP block, and write over it
* with a copy of the last-synced uberblock, whose timestamp
* has been updated to reflect that the pool is in use.
*/
static void
mmp_write_uberblock(spa_t *spa)
{
int flags = ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_CANFAIL;
mmp_thread_t *mmp = &spa->spa_mmp;
uberblock_t *ub;
vdev_t *vd;
int label;
uint64_t offset;
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
vd = mmp_random_leaf(spa->spa_root_vdev);
if (vd == NULL) {
spa_config_exit(spa, SCL_STATE, FTAG);
return;
}
mutex_enter(&mmp->mmp_io_lock);
if (mmp->mmp_zio_root == NULL)
mmp->mmp_zio_root = zio_root(spa, NULL, NULL,
flags | ZIO_FLAG_GODFATHER);
ub = &mmp->mmp_ub;
ub->ub_timestamp = gethrestime_sec();
ub->ub_mmp_magic = MMP_MAGIC;
ub->ub_mmp_delay = mmp->mmp_delay;
vd->vdev_mmp_pending = gethrtime();
zio_t *zio = zio_null(mmp->mmp_zio_root, spa, NULL, NULL, NULL, flags);
abd_t *ub_abd = abd_alloc_for_io(VDEV_UBERBLOCK_SIZE(vd), B_TRUE);
abd_zero(ub_abd, VDEV_UBERBLOCK_SIZE(vd));
abd_copy_from_buf(ub_abd, ub, sizeof (uberblock_t));
mutex_exit(&mmp->mmp_io_lock);
offset = VDEV_UBERBLOCK_OFFSET(vd, VDEV_UBERBLOCK_COUNT(vd) -
MMP_BLOCKS_PER_LABEL + spa_get_random(MMP_BLOCKS_PER_LABEL));
label = spa_get_random(VDEV_LABELS);
vdev_label_write(zio, vd, label, ub_abd, offset,
VDEV_UBERBLOCK_SIZE(vd), mmp_write_done, mmp,
flags | ZIO_FLAG_DONT_PROPAGATE);
spa_mmp_history_add(ub->ub_txg, ub->ub_timestamp, ub->ub_mmp_delay, vd,
label);
zio_nowait(zio);
}
static void
spa_load_stats_update(spa_t *spa)
{
int i, nnormal = 0, nspecial = 0;
vdev_t *rvd;
spa_special_stat_t spa_stat = {0};
/*
* walk the top level vdevs and calculate average stats for
* the normal and special classes
*/
spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
rvd = spa->spa_root_vdev;
ASSERT(rvd);
for (i = 0; i < rvd->vdev_children; i++) {
vdev_t *vd = rvd->vdev_child[i];
ASSERT(vd);
if (vd->vdev_islog || vd->vdev_ishole ||
vd->vdev_isspare || vd->vdev_isl2cache)
continue;
if (vd->vdev_isspecial) {
spa_vdev_walk_stats(vd, &spa_stat.special_lt,
&spa_stat.special_ut, &nspecial);
} else {
spa_vdev_walk_stats(vd, &spa_stat.normal_lt,
&spa_stat.normal_ut, &nnormal);
}
}
spa_config_exit(spa, SCL_VDEV, FTAG);
DTRACE_PROBE4(spa_vdev_stats, int, nnormal, int, nspecial,
int, spa_stat.normal_ut, int, spa_stat.special_ut);
/* calculate averages */
if (nnormal) {
spa->spa_special_stat.normal_lt = spa_stat.normal_lt/nnormal;
spa->spa_special_stat.normal_ut = spa_stat.normal_ut/nnormal;
}
if (nspecial) {
spa->spa_special_stat.special_lt = spa_stat.special_lt/nspecial;
spa->spa_special_stat.special_ut = spa_stat.special_ut/nspecial;
}
}
/*
* Update all disk labels, generate a fresh config based on the current
* in-core state, and sync the global config cache (do not sync the config
* cache if this is a booting rootpool).
*/
void
spa_config_update(spa_t *spa, int what)
{
vdev_t *rvd = spa->spa_root_vdev;
uint64_t txg;
int c;
ASSERT(MUTEX_HELD(&spa_namespace_lock));
spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
txg = spa_last_synced_txg(spa) + 1;
if (what == SPA_CONFIG_UPDATE_POOL) {
vdev_config_dirty(rvd);
} else {
/*
* If we have top-level vdevs that were added but have
* not yet been prepared for allocation, do that now.
* (It's safe now because the config cache is up to date,
* so it will be able to translate the new DVAs.)
* See comments in spa_vdev_add() for full details.
*/
for (c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
if (tvd->vdev_ms_array == 0)
vdev_metaslab_set_size(tvd);
vdev_expand(tvd, txg);
}
}
spa_config_exit(spa, SCL_ALL, FTAG);
/*
* Wait for the mosconfig to be regenerated and synced.
*/
txg_wait_synced(spa->spa_dsl_pool, txg);
/*
* Update the global config cache to reflect the new mosconfig.
*/
if (!spa->spa_is_root) {
spa_write_cachefile(spa, B_FALSE,
what != SPA_CONFIG_UPDATE_POOL);
}
if (what == SPA_CONFIG_UPDATE_POOL)
spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS);
}
void
zil_flush_vdevs(zilog_t *zilog)
{
spa_t *spa = zilog->zl_spa;
avl_tree_t *t = &zilog->zl_vdev_tree;
void *cookie = NULL;
zil_vdev_node_t *zv;
zio_t *zio;
ASSERT(zilog->zl_writer);
/*
* We don't need zl_vdev_lock here because we're the zl_writer,
* and all zl_get_data() callbacks are done.
*/
if (avl_numnodes(t) == 0)
return;
spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
while ((zv = avl_destroy_nodes(t, &cookie)) != NULL) {
vdev_t *vd = vdev_lookup_top(spa, zv->zv_vdev);
if (vd != NULL)
zio_flush(zio, vd);
kmem_free(zv, sizeof (*zv));
}
/*
* Wait for all the flushes to complete. Not all devices actually
* support the DKIOCFLUSHWRITECACHE ioctl, so it's OK if it fails.
*/
(void) zio_wait(zio);
spa_config_exit(spa, SCL_STATE, FTAG);
}
/*
* Generate the pool's configuration based on the current in-core state.
*
* We infer whether to generate a complete config or just one top-level config
* based on whether vd is the root vdev.
*/
nvlist_t *
spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats)
{
nvlist_t *config, *nvroot;
vdev_t *rvd = spa->spa_root_vdev;
unsigned long hostid = 0;
boolean_t locked = B_FALSE;
uint64_t split_guid;
if (vd == NULL) {
vd = rvd;
locked = B_TRUE;
spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
}
ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) ==
(SCL_CONFIG | SCL_STATE));
/*
* If txg is -1, report the current value of spa->spa_config_txg.
*/
if (txg == -1ULL)
txg = spa->spa_config_txg;
VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
spa_version(spa)) == 0);
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
spa_name(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
spa_state(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
txg) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
spa_guid(spa)) == 0);
VERIFY(spa->spa_comment == NULL || nvlist_add_string(config,
ZPOOL_CONFIG_COMMENT, spa->spa_comment) == 0);
#ifdef _KERNEL
hostid = zone_get_hostid(NULL);
#else /* _KERNEL */
/*
* We're emulating the system's hostid in userland, so we can't use
* zone_get_hostid().
*/
(void) ddi_strtoul(hw_serial, NULL, 10, &hostid);
#endif /* _KERNEL */
if (hostid != 0) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
hostid) == 0);
}
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
utsname.nodename) == 0);
if (vd != rvd) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID,
vd->vdev_top->vdev_guid) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID,
vd->vdev_guid) == 0);
if (vd->vdev_isspare)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE,
1ULL) == 0);
if (vd->vdev_islog)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG,
1ULL) == 0);
vd = vd->vdev_top; /* label contains top config */
} else {
/*
* Only add the (potentially large) split information
* in the mos config, and not in the vdev labels
*/
if (spa->spa_config_splitting != NULL)
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT,
spa->spa_config_splitting) == 0);
}
/*
* Add the top-level config. We even add this on pools which
* don't support holes in the namespace.
*/
vdev_top_config_generate(spa, config);
/*
* If we're splitting, record the original pool's guid.
*/
if (spa->spa_config_splitting != NULL &&
nvlist_lookup_uint64(spa->spa_config_splitting,
ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID,
split_guid) == 0);
}
nvroot = vdev_config_generate(spa, vd, getstats, 0);
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
nvlist_free(nvroot);
/*
* Store what's necessary for reading the MOS in the label.
*/
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ,
spa->spa_label_features) == 0);
if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) {
ddt_histogram_t *ddh;
ddt_stat_t *dds;
ddt_object_t *ddo;
ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_SLEEP);
ddt_get_dedup_histogram(spa, ddh);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_HISTOGRAM,
(uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)) == 0);
kmem_free(ddh, sizeof (ddt_histogram_t));
ddo = kmem_zalloc(sizeof (ddt_object_t), KM_SLEEP);
ddt_get_dedup_object_stats(spa, ddo);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_OBJ_STATS,
(uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)) == 0);
kmem_free(ddo, sizeof (ddt_object_t));
dds = kmem_zalloc(sizeof (ddt_stat_t), KM_SLEEP);
ddt_get_dedup_stats(spa, dds);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_STATS,
(uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)) == 0);
kmem_free(dds, sizeof (ddt_stat_t));
}
if (locked)
spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
return (config);
}
/*
* Generate the pool's configuration based on the current in-core state.
*
* We infer whether to generate a complete config or just one top-level config
* based on whether vd is the root vdev.
*/
nvlist_t *
spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats)
{
nvlist_t *config, *nvroot;
vdev_t *rvd = spa->spa_root_vdev;
unsigned long hostid = 0;
boolean_t locked = B_FALSE;
uint64_t split_guid;
char *pool_name;
if (vd == NULL) {
vd = rvd;
locked = B_TRUE;
spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
}
ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) ==
(SCL_CONFIG | SCL_STATE));
/*
* If txg is -1, report the current value of spa->spa_config_txg.
*/
if (txg == -1ULL)
txg = spa->spa_config_txg;
/*
* Originally, users had to handle spa namespace collisions by either
* exporting the already imported pool or by specifying a new name for
* the pool with a conflicting name. In the case of root pools from
* virtual guests, neither approach to collision resolution is
* reasonable. This is addressed by extending the new name syntax with
* an option to specify that the new name is temporary. When specified,
* ZFS_IMPORT_TEMP_NAME will be set in spa->spa_import_flags to tell us
* to use the previous name, which we do below.
*/
if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) {
VERIFY0(nvlist_lookup_string(spa->spa_config,
ZPOOL_CONFIG_POOL_NAME, &pool_name));
} else
pool_name = spa_name(spa);
VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
spa_version(spa)) == 0);
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
pool_name) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
spa_state(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
txg) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
spa_guid(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA,
spa->spa_errata) == 0);
VERIFY(spa->spa_comment == NULL || nvlist_add_string(config,
ZPOOL_CONFIG_COMMENT, spa->spa_comment) == 0);
#ifdef _KERNEL
hostid = zone_get_hostid(NULL);
#else /* _KERNEL */
/*
* We're emulating the system's hostid in userland, so we can't use
* zone_get_hostid().
*/
(void) ddi_strtoul(hw_serial, NULL, 10, &hostid);
#endif /* _KERNEL */
if (hostid != 0) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
hostid) == 0);
}
VERIFY0(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
utsname()->nodename));
if (vd != rvd) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID,
vd->vdev_top->vdev_guid) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID,
vd->vdev_guid) == 0);
if (vd->vdev_isspare)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE,
1ULL) == 0);
if (vd->vdev_islog)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG,
1ULL) == 0);
vd = vd->vdev_top; /* label contains top config */
} else {
/*
* Only add the (potentially large) split information
* in the mos config, and not in the vdev labels
*/
if (spa->spa_config_splitting != NULL)
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT,
spa->spa_config_splitting) == 0);
}
/*
* Add the top-level config. We even add this on pools which
* don't support holes in the namespace.
*/
vdev_top_config_generate(spa, config);
/*
* If we're splitting, record the original pool's guid.
*/
if (spa->spa_config_splitting != NULL &&
nvlist_lookup_uint64(spa->spa_config_splitting,
ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID,
split_guid) == 0);
}
nvroot = vdev_config_generate(spa, vd, getstats, 0);
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
nvlist_free(nvroot);
/*
* Store what's necessary for reading the MOS in the label.
*/
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ,
spa->spa_label_features) == 0);
if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) {
ddt_histogram_t *ddh;
ddt_stat_t *dds;
ddt_object_t *ddo;
ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_SLEEP);
ddt_get_dedup_histogram(spa, ddh);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_HISTOGRAM,
(uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)) == 0);
kmem_free(ddh, sizeof (ddt_histogram_t));
ddo = kmem_zalloc(sizeof (ddt_object_t), KM_SLEEP);
ddt_get_dedup_object_stats(spa, ddo);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_OBJ_STATS,
(uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)) == 0);
kmem_free(ddo, sizeof (ddt_object_t));
dds = kmem_zalloc(sizeof (ddt_stat_t), KM_SLEEP);
ddt_get_dedup_stats(spa, dds);
VERIFY(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_DDT_STATS,
(uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)) == 0);
kmem_free(dds, sizeof (ddt_stat_t));
}
if (locked)
spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
return (config);
}
static int
zfs_vfs_getattr(struct mount *mp, struct vfs_attr *fsap, __unused vfs_context_t context)
{
zfsvfs_t *zfsvfs = vfs_fsprivate(mp);
uint64_t refdbytes, availbytes, usedobjs, availobjs;
ZFS_ENTER(zfsvfs);
dmu_objset_space(zfsvfs->z_os,
&refdbytes, &availbytes, &usedobjs, &availobjs);
VFSATTR_RETURN(fsap, f_objcount, usedobjs);
VFSATTR_RETURN(fsap, f_maxobjcount, 0x7fffffffffffffff);
/*
* Carbon depends on f_filecount and f_dircount so
* make up some values based on total objects.
*/
VFSATTR_RETURN(fsap, f_filecount, usedobjs - (usedobjs / 4));
VFSATTR_RETURN(fsap, f_dircount, usedobjs / 4);
/*
* The underlying storage pool actually uses multiple block sizes.
* We report the fragsize as the smallest block size we support,
* and we report our blocksize as the filesystem's maximum blocksize.
*/
VFSATTR_RETURN(fsap, f_bsize, 1UL << SPA_MINBLOCKSHIFT);
VFSATTR_RETURN(fsap, f_iosize, zfsvfs->z_max_blksz);
/*
* The following report "total" blocks of various kinds in the
* file system, but reported in terms of f_frsize - the
* "fragment" size.
*/
VFSATTR_RETURN(fsap, f_blocks,
(u_int64_t)((refdbytes + availbytes) >> SPA_MINBLOCKSHIFT));
VFSATTR_RETURN(fsap, f_bfree, (u_int64_t)(availbytes >> SPA_MINBLOCKSHIFT));
VFSATTR_RETURN(fsap, f_bavail, fsap->f_bfree); /* no root reservation */
VFSATTR_RETURN(fsap, f_bused, fsap->f_blocks - fsap->f_bfree);
/*
* statvfs() should really be called statufs(), because it assumes
* static metadata. ZFS doesn't preallocate files, so the best
* we can do is report the max that could possibly fit in f_files,
* and that minus the number actually used in f_ffree.
* For f_ffree, report the smaller of the number of object available
* and the number of blocks (each object will take at least a block).
*/
VFSATTR_RETURN(fsap, f_ffree, (u_int64_t)MIN(availobjs, fsap->f_bfree));
VFSATTR_RETURN(fsap, f_files, fsap->f_ffree + usedobjs);
#if 0
statp->f_flag = vf_to_stf(vfsp->vfs_flag);
#endif
if (VFSATTR_IS_ACTIVE(fsap, f_fsid)) {
VFSATTR_RETURN(fsap, f_fsid, vfs_statfs(mp)->f_fsid);
}
if (VFSATTR_IS_ACTIVE(fsap, f_capabilities)) {
bcopy(&zfs_capabilities, &fsap->f_capabilities, sizeof (zfs_capabilities));
VFSATTR_SET_SUPPORTED(fsap, f_capabilities);
}
if (VFSATTR_IS_ACTIVE(fsap, f_attributes)) {
bcopy(&zfs_attributes, &fsap->f_attributes.validattr, sizeof (zfs_attributes));
bcopy(&zfs_attributes, &fsap->f_attributes.nativeattr, sizeof (zfs_attributes));
VFSATTR_SET_SUPPORTED(fsap, f_attributes);
}
if (VFSATTR_IS_ACTIVE(fsap, f_create_time)) {
dmu_objset_stats_t dmu_stat;
dmu_objset_fast_stat(zfsvfs->z_os, &dmu_stat);
fsap->f_create_time.tv_sec = dmu_stat.dds_creation_time;
fsap->f_create_time.tv_nsec = 0;
VFSATTR_SET_SUPPORTED(fsap, f_create_time);
}
if (VFSATTR_IS_ACTIVE(fsap, f_modify_time)) {
if (zfsvfs->z_mtime_vp != NULL) {
znode_t *mzp;
mzp = VTOZ(zfsvfs->z_mtime_vp);
ZFS_TIME_DECODE(&fsap->f_modify_time, mzp->z_phys->zp_mtime);
} else {
fsap->f_modify_time.tv_sec = 0;
fsap->f_modify_time.tv_nsec = 0;
}
VFSATTR_SET_SUPPORTED(fsap, f_modify_time);
}
/*
* For Carbon compatibility, pretend to support this legacy/unused
* attribute
*/
if (VFSATTR_IS_ACTIVE(fsap, f_backup_time)) {
fsap->f_backup_time.tv_sec = 0;
fsap->f_backup_time.tv_nsec = 0;
VFSATTR_SET_SUPPORTED(fsap, f_backup_time);
}
if (VFSATTR_IS_ACTIVE(fsap, f_vol_name)) {
spa_t *spa = dmu_objset_spa(zfsvfs->z_os);
spa_config_enter(spa, RW_READER, FTAG);
strlcpy(fsap->f_vol_name, spa_name(spa), MAXPATHLEN);
spa_config_exit(spa, FTAG);
VFSATTR_SET_SUPPORTED(fsap, f_vol_name);
}
VFSATTR_RETURN(fsap, f_fssubtype, 0);
VFSATTR_RETURN(fsap, f_signature, 0x5a21); /* 'Z!' */
VFSATTR_RETURN(fsap, f_carbon_fsid, 0);
ZFS_EXIT(zfsvfs);
return (0);
}
/* ARGSUSED */
static int
spa_condense_indirect_thread(void *arg, zthr_t *zthr)
{
spa_t *spa = arg;
vdev_t *vd;
ASSERT3P(spa->spa_condensing_indirect, !=, NULL);
spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
vd = vdev_lookup_top(spa, spa->spa_condensing_indirect_phys.scip_vdev);
ASSERT3P(vd, !=, NULL);
spa_config_exit(spa, SCL_VDEV, FTAG);
spa_condensing_indirect_t *sci = spa->spa_condensing_indirect;
spa_condensing_indirect_phys_t *scip =
&spa->spa_condensing_indirect_phys;
uint32_t *counts;
uint64_t start_index;
vdev_indirect_mapping_t *old_mapping = vd->vdev_indirect_mapping;
space_map_t *prev_obsolete_sm = NULL;
ASSERT3U(vd->vdev_id, ==, scip->scip_vdev);
ASSERT(scip->scip_next_mapping_object != 0);
ASSERT(scip->scip_prev_obsolete_sm_object != 0);
ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);
for (int i = 0; i < TXG_SIZE; i++) {
/*
* The list must start out empty in order for the
* _commit_sync() sync task to be properly registered
* on the first call to _commit_entry(); so it's wise
* to double check and ensure we actually are starting
* with empty lists.
*/
ASSERT(list_is_empty(&sci->sci_new_mapping_entries[i]));
}
VERIFY0(space_map_open(&prev_obsolete_sm, spa->spa_meta_objset,
scip->scip_prev_obsolete_sm_object, 0, vd->vdev_asize, 0));
space_map_update(prev_obsolete_sm);
counts = vdev_indirect_mapping_load_obsolete_counts(old_mapping);
if (prev_obsolete_sm != NULL) {
vdev_indirect_mapping_load_obsolete_spacemap(old_mapping,
counts, prev_obsolete_sm);
}
space_map_close(prev_obsolete_sm);
/*
* Generate new mapping. Determine what index to continue from
* based on the max offset that we've already written in the
* new mapping.
*/
uint64_t max_offset =
vdev_indirect_mapping_max_offset(sci->sci_new_mapping);
if (max_offset == 0) {
/* We haven't written anything to the new mapping yet. */
start_index = 0;
} else {
/*
* Pick up from where we left off. _entry_for_offset()
* returns a pointer into the vim_entries array. If
* max_offset is greater than any of the mappings
* contained in the table NULL will be returned and
* that indicates we've exhausted our iteration of the
* old_mapping.
*/
vdev_indirect_mapping_entry_phys_t *entry =
vdev_indirect_mapping_entry_for_offset_or_next(old_mapping,
max_offset);
if (entry == NULL) {
/*
* We've already written the whole new mapping.
* This special value will cause us to skip the
* generate_new_mapping step and just do the sync
* task to complete the condense.
*/
start_index = UINT64_MAX;
} else {
start_index = entry - old_mapping->vim_entries;
ASSERT3U(start_index, <,
vdev_indirect_mapping_num_entries(old_mapping));
}
}
spa_condense_indirect_generate_new_mapping(vd, counts,
start_index, zthr);
vdev_indirect_mapping_free_obsolete_counts(old_mapping, counts);
/*
* If the zthr has received a cancellation signal while running
* in generate_new_mapping() or at any point after that, then bail
* early. We don't want to complete the condense if the spa is
* shutting down.
*/
if (zthr_iscancelled(zthr))
return (0);
VERIFY0(dsl_sync_task(spa_name(spa), NULL,
spa_condense_indirect_complete_sync, sci, 0,
ZFS_SPACE_CHECK_EXTRA_RESERVED));
return (0);
thread_exit();
}
static void
txg_sync_thread(dsl_pool_t *dp)
{
spa_t *spa = dp->dp_spa;
tx_state_t *tx = &dp->dp_tx;
callb_cpr_t cpr;
vdev_stat_t *vs1, *vs2;
clock_t start, delta;
#ifdef _KERNEL
/*
* Annotate this process with a flag that indicates that it is
* unsafe to use KM_SLEEP during memory allocations due to the
* potential for a deadlock. KM_PUSHPAGE should be used instead.
*/
current->flags |= PF_NOFS;
#endif /* _KERNEL */
txg_thread_enter(tx, &cpr);
vs1 = kmem_alloc(sizeof (vdev_stat_t), KM_PUSHPAGE);
vs2 = kmem_alloc(sizeof (vdev_stat_t), KM_PUSHPAGE);
start = delta = 0;
for (;;) {
clock_t timer, timeout;
uint64_t txg;
uint64_t ndirty;
timeout = zfs_txg_timeout * hz;
/*
* We sync when we're scanning, there's someone waiting
* on us, or the quiesce thread has handed off a txg to
* us, or we have reached our timeout.
*/
timer = (delta >= timeout ? 0 : timeout - delta);
while (!dsl_scan_active(dp->dp_scan) &&
!tx->tx_exiting && timer > 0 &&
tx->tx_synced_txg >= tx->tx_sync_txg_waiting &&
tx->tx_quiesced_txg == 0 &&
dp->dp_dirty_total < zfs_dirty_data_sync) {
dprintf("waiting; tx_synced=%llu waiting=%llu dp=%p\n",
tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp);
txg_thread_wait(tx, &cpr, &tx->tx_sync_more_cv, timer);
delta = ddi_get_lbolt() - start;
timer = (delta > timeout ? 0 : timeout - delta);
}
/*
* Wait until the quiesce thread hands off a txg to us,
* prompting it to do so if necessary.
*/
while (!tx->tx_exiting && tx->tx_quiesced_txg == 0) {
if (tx->tx_quiesce_txg_waiting < tx->tx_open_txg+1)
tx->tx_quiesce_txg_waiting = tx->tx_open_txg+1;
cv_broadcast(&tx->tx_quiesce_more_cv);
txg_thread_wait(tx, &cpr, &tx->tx_quiesce_done_cv, 0);
}
if (tx->tx_exiting) {
kmem_free(vs2, sizeof (vdev_stat_t));
kmem_free(vs1, sizeof (vdev_stat_t));
txg_thread_exit(tx, &cpr, &tx->tx_sync_thread);
}
spa_config_enter(spa, SCL_ALL, FTAG, RW_READER);
vdev_get_stats(spa->spa_root_vdev, vs1);
spa_config_exit(spa, SCL_ALL, FTAG);
/*
* Consume the quiesced txg which has been handed off to
* us. This may cause the quiescing thread to now be
* able to quiesce another txg, so we must signal it.
*/
txg = tx->tx_quiesced_txg;
tx->tx_quiesced_txg = 0;
tx->tx_syncing_txg = txg;
DTRACE_PROBE2(txg__syncing, dsl_pool_t *, dp, uint64_t, txg);
cv_broadcast(&tx->tx_quiesce_more_cv);
dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",
txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting);
mutex_exit(&tx->tx_sync_lock);
spa_txg_history_set(spa, txg, TXG_STATE_WAIT_FOR_SYNC,
gethrtime());
ndirty = dp->dp_dirty_pertxg[txg & TXG_MASK];
start = ddi_get_lbolt();
spa_sync(spa, txg);
delta = ddi_get_lbolt() - start;
mutex_enter(&tx->tx_sync_lock);
tx->tx_synced_txg = txg;
tx->tx_syncing_txg = 0;
DTRACE_PROBE2(txg__synced, dsl_pool_t *, dp, uint64_t, txg);
cv_broadcast(&tx->tx_sync_done_cv);
/*
* Dispatch commit callbacks to worker threads.
*/
txg_dispatch_callbacks(dp, txg);
spa_config_enter(spa, SCL_ALL, FTAG, RW_READER);
vdev_get_stats(spa->spa_root_vdev, vs2);
spa_config_exit(spa, SCL_ALL, FTAG);
spa_txg_history_set_io(spa, txg,
vs2->vs_bytes[ZIO_TYPE_READ]-vs1->vs_bytes[ZIO_TYPE_READ],
vs2->vs_bytes[ZIO_TYPE_WRITE]-vs1->vs_bytes[ZIO_TYPE_WRITE],
vs2->vs_ops[ZIO_TYPE_READ]-vs1->vs_ops[ZIO_TYPE_READ],
vs2->vs_ops[ZIO_TYPE_WRITE]-vs1->vs_ops[ZIO_TYPE_WRITE],
ndirty);
spa_txg_history_set(spa, txg, TXG_STATE_SYNCED, gethrtime());
}
}
/*
* Generate the pool's configuration based on the current in-core state.
* We infer whether to generate a complete config or just one top-level config
* based on whether vd is the root vdev.
*/
nvlist_t *
spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats)
{
nvlist_t *config, *nvroot;
vdev_t *rvd = spa->spa_root_vdev;
unsigned long hostid = 0;
boolean_t locked = B_FALSE;
if (vd == NULL) {
vd = rvd;
locked = B_TRUE;
spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER);
}
ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) ==
(SCL_CONFIG | SCL_STATE));
/*
* If txg is -1, report the current value of spa->spa_config_txg.
*/
if (txg == -1ULL)
txg = spa->spa_config_txg;
VERIFY(nvlist_alloc(&config, NV_UNIQUE_NAME, KM_SLEEP) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
spa_version(spa)) == 0);
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME,
spa_name(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
spa_state(spa)) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG,
txg) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
spa_guid(spa)) == 0);
#ifdef _KERNEL
hostid = zone_get_hostid(NULL);
#else /* _KERNEL */
/*
* We're emulating the system's hostid in userland, so we can't use
* zone_get_hostid().
*/
(void) ddi_strtoul(hw_serial, NULL, 10, &hostid);
#endif /* _KERNEL */
if (hostid != 0) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
hostid) == 0);
}
VERIFY(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
utsname.nodename) == 0);
if (vd != rvd) {
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID,
vd->vdev_top->vdev_guid) == 0);
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_GUID,
vd->vdev_guid) == 0);
if (vd->vdev_isspare)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE,
1ULL) == 0);
if (vd->vdev_islog)
VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG,
1ULL) == 0);
vd = vd->vdev_top; /* label contains top config */
}
nvroot = vdev_config_generate(spa, vd, getstats, B_FALSE, B_FALSE);
VERIFY(nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot) == 0);
nvlist_free(nvroot);
if (locked)
spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG);
return (config);
}
