Native Encryption for ZFS on Linux

This change incorporates three major pieces:

The first change is a keystore that manages wrapping
and encryption keys for encrypted datasets. These
commands mostly involve manipulating the new
DSL Crypto Key ZAP Objects that live in the MOS. Each
encrypted dataset has its own DSL Crypto Key that is
protected with a user's key. This level of indirection
allows users to change their keys without re-encrypting
their entire datasets. The change implements the new
subcommands "zfs load-key", "zfs unload-key" and
"zfs change-key" which allow the user to manage their
encryption keys and settings. In addition, several new
flags and properties have been added to allow dataset
creation and to make mounting and unmounting more
convenient.

The second piece of this patch provides the ability to
encrypt, decyrpt, and authenticate protected datasets.
Each object set maintains a Merkel tree of Message
Authentication Codes that protect the lower layers,
similarly to how checksums are maintained. This part
impacts the zio layer, which handles the actual
encryption and generation of MACs, as well as the ARC
and DMU, which need to be able to handle encrypted
buffers and protected data.

The last addition is the ability to do raw, encrypted
sends and receives. The idea here is to send raw
encrypted and compressed data and receive it exactly
as is on a backup system. This means that the dataset
on the receiving system is protected using the same
user key that is in use on the sending side. By doing
so, datasets can be efficiently backed up to an
untrusted system without fear of data being
compromised.

Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Signed-off-by: Tom Caputi <tcaputi@datto.com>
Closes #494 
Closes #5769
This commit is contained in:
Tom Caputi
2017-08-14 13:36:48 -04:00
committed by Brian Behlendorf
parent 376994828f
commit b525630342
163 changed files with 16091 additions and 1204 deletions
+66 -19
View File
@@ -1169,6 +1169,8 @@ spa_activate(spa_t *spa, int mode)
spa_error_entry_compare, sizeof (spa_error_entry_t),
offsetof(spa_error_entry_t, se_avl));
spa_keystore_init(&spa->spa_keystore);
/*
* This taskq is used to perform zvol-minor-related tasks
* asynchronously. This has several advantages, including easy
@@ -1246,10 +1248,11 @@ spa_deactivate(spa_t *spa)
* still have errors left in the queues. Empty them just in case.
*/
spa_errlog_drain(spa);
avl_destroy(&spa->spa_errlist_scrub);
avl_destroy(&spa->spa_errlist_last);
spa_keystore_fini(&spa->spa_keystore);
spa->spa_state = POOL_STATE_UNINITIALIZED;
mutex_enter(&spa->spa_proc_lock);
@@ -2094,8 +2097,8 @@ spa_load_verify(spa_t *spa)
if (spa_load_verify_metadata) {
error = traverse_pool(spa, spa->spa_verify_min_txg,
TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA,
spa_load_verify_cb, rio);
TRAVERSE_PRE | TRAVERSE_PREFETCH_METADATA |
TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio);
}
(void) zio_wait(rio);
@@ -2301,7 +2304,7 @@ spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type,
spa->spa_loaded_ts.tv_nsec = 0;
}
if (error != EBADF) {
zfs_ereport_post(ereport, spa, NULL, NULL, 0, 0);
zfs_ereport_post(ereport, spa, NULL, NULL, NULL, 0, 0);
}
}
spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE;
@@ -3978,12 +3981,28 @@ spa_l2cache_drop(spa_t *spa)
}
}
/*
* Verify encryption parameters for spa creation. If we are encrypting, we must
* have the encryption feature flag enabled.
*/
static int
spa_create_check_encryption_params(dsl_crypto_params_t *dcp,
boolean_t has_encryption)
{
if (dcp->cp_crypt != ZIO_CRYPT_OFF &&
dcp->cp_crypt != ZIO_CRYPT_INHERIT &&
!has_encryption)
return (SET_ERROR(ENOTSUP));
return (dmu_objset_create_crypt_check(NULL, dcp));
}
/*
* Pool Creation
*/
int
spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
nvlist_t *zplprops)
nvlist_t *zplprops, dsl_crypto_params_t *dcp)
{
spa_t *spa;
char *altroot = NULL;
@@ -3994,8 +4013,11 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
uint64_t txg = TXG_INITIAL;
nvlist_t **spares, **l2cache;
uint_t nspares, nl2cache;
uint64_t version, obj;
uint64_t version, obj, root_dsobj = 0;
boolean_t has_features;
boolean_t has_encryption;
spa_feature_t feat;
char *feat_name;
nvpair_t *elem;
int c, i;
char *poolname;
@@ -4038,10 +4060,28 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
spa->spa_import_flags |= ZFS_IMPORT_TEMP_NAME;
has_features = B_FALSE;
has_encryption = B_FALSE;
for (elem = nvlist_next_nvpair(props, NULL);
elem != NULL; elem = nvlist_next_nvpair(props, elem)) {
if (zpool_prop_feature(nvpair_name(elem)))
if (zpool_prop_feature(nvpair_name(elem))) {
has_features = B_TRUE;
feat_name = strchr(nvpair_name(elem), '@') + 1;
VERIFY0(zfeature_lookup_name(feat_name, &feat));
if (feat == SPA_FEATURE_ENCRYPTION)
has_encryption = B_TRUE;
}
}
/* verify encryption params, if they were provided */
if (dcp != NULL) {
error = spa_create_check_encryption_params(dcp, has_encryption);
if (error != 0) {
spa_deactivate(spa);
spa_remove(spa);
mutex_exit(&spa_namespace_lock);
return (error);
}
}
if (has_features || nvlist_lookup_uint64(props,
@@ -4131,8 +4171,7 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
}
spa->spa_is_initializing = B_TRUE;
spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, txg);
spa->spa_meta_objset = dp->dp_meta_objset;
spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, dcp, txg);
spa->spa_is_initializing = B_FALSE;
/*
@@ -4157,9 +4196,6 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
cmn_err(CE_PANIC, "failed to add pool config");
}
if (spa_version(spa) >= SPA_VERSION_FEATURES)
spa_feature_create_zap_objects(spa, tx);
if (zap_add(spa->spa_meta_objset,
DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION,
sizeof (uint64_t), 1, &version, tx) != 0) {
@@ -4220,15 +4256,26 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_t *props,
dmu_tx_commit(tx);
spa->spa_sync_on = B_TRUE;
txg_sync_start(spa->spa_dsl_pool);
mmp_thread_start(spa);
/*
* We explicitly wait for the first transaction to complete so that our
* bean counters are appropriately updated.
* If the root dataset is encrypted we will need to create key mappings
* for the zio layer before we start to write any data to disk and hold
* them until after the first txg has been synced. Waiting for the first
* transaction to complete also ensures that our bean counters are
* appropriately updated.
*/
txg_wait_synced(spa->spa_dsl_pool, txg);
if (dp->dp_root_dir->dd_crypto_obj != 0) {
root_dsobj = dsl_dir_phys(dp->dp_root_dir)->dd_head_dataset_obj;
VERIFY0(spa_keystore_create_mapping_impl(spa, root_dsobj,
dp->dp_root_dir, FTAG));
}
spa->spa_sync_on = B_TRUE;
txg_sync_start(dp);
mmp_thread_start(spa);
txg_wait_synced(dp, txg);
if (dp->dp_root_dir->dd_crypto_obj != 0)
VERIFY0(spa_keystore_remove_mapping(spa, root_dsobj, FTAG));
spa_config_sync(spa, B_FALSE, B_TRUE);
spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_CREATE);