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.
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.
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]
23 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Portions Copyright 2011 Martin Matuska
25 * Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved.
26 * Portions Copyright 2012 Pawel Jakub Dawidek <pawel@dawidek.net>
27 * Copyright (c) 2014, 2016 Joyent, Inc. All rights reserved.
28 * Copyright 2016 Nexenta Systems, Inc. All rights reserved.
29 * Copyright (c) 2014, Joyent, Inc. All rights reserved.
30 * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
31 * Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
32 * Copyright (c) 2013 Steven Hartland. All rights reserved.
33 * Copyright (c) 2014 Integros [integros.com]
34 * Copyright 2016 Toomas Soome <tsoome@me.com>
35 * Copyright (c) 2016 Actifio, Inc. All rights reserved.
36 * Copyright (c) 2018, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
37 * Copyright 2017 RackTop Systems.
38 * Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
39 * Copyright (c) 2019 Datto Inc.
45 * This file handles the ioctls to /dev/zfs, used for configuring ZFS storage
46 * pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool.
48 * There are two ways that we handle ioctls: the legacy way where almost
49 * all of the logic is in the ioctl callback, and the new way where most
50 * of the marshalling is handled in the common entry point, zfsdev_ioctl().
52 * Non-legacy ioctls should be registered by calling
53 * zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked
54 * from userland by lzc_ioctl().
56 * The registration arguments are as follows:
59 * The name of the ioctl. This is used for history logging. If the
60 * ioctl returns successfully (the callback returns 0), and allow_log
61 * is true, then a history log entry will be recorded with the input &
62 * output nvlists. The log entry can be printed with "zpool history -i".
65 * The ioctl request number, which userland will pass to ioctl(2).
66 * We want newer versions of libzfs and libzfs_core to run against
67 * existing zfs kernel modules (i.e. a deferred reboot after an update).
68 * Therefore the ioctl numbers cannot change from release to release.
70 * zfs_secpolicy_func_t *secpolicy
71 * This function will be called before the zfs_ioc_func_t, to
72 * determine if this operation is permitted. It should return EPERM
73 * on failure, and 0 on success. Checks include determining if the
74 * dataset is visible in this zone, and if the user has either all
75 * zfs privileges in the zone (SYS_MOUNT), or has been granted permission
76 * to do this operation on this dataset with "zfs allow".
78 * zfs_ioc_namecheck_t namecheck
79 * This specifies what to expect in the zfs_cmd_t:zc_name -- a pool
80 * name, a dataset name, or nothing. If the name is not well-formed,
81 * the ioctl will fail and the callback will not be called.
82 * Therefore, the callback can assume that the name is well-formed
83 * (e.g. is null-terminated, doesn't have more than one '@' character,
84 * doesn't have invalid characters).
86 * zfs_ioc_poolcheck_t pool_check
87 * This specifies requirements on the pool state. If the pool does
88 * not meet them (is suspended or is readonly), the ioctl will fail
89 * and the callback will not be called. If any checks are specified
90 * (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME.
91 * Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED |
92 * POOL_CHECK_READONLY).
94 * zfs_ioc_key_t *nvl_keys
95 * The list of expected/allowable innvl input keys. This list is used
96 * to validate the nvlist input to the ioctl.
98 * boolean_t smush_outnvlist
99 * If smush_outnvlist is true, then the output is presumed to be a
100 * list of errors, and it will be "smushed" down to fit into the
101 * caller's buffer, by removing some entries and replacing them with a
102 * single "N_MORE_ERRORS" entry indicating how many were removed. See
103 * nvlist_smush() for details. If smush_outnvlist is false, and the
104 * outnvlist does not fit into the userland-provided buffer, then the
105 * ioctl will fail with ENOMEM.
107 * zfs_ioc_func_t *func
108 * The callback function that will perform the operation.
110 * The callback should return 0 on success, or an error number on
111 * failure. If the function fails, the userland ioctl will return -1,
112 * and errno will be set to the callback's return value. The callback
113 * will be called with the following arguments:
116 * The name of the pool or dataset to operate on, from
117 * zfs_cmd_t:zc_name. The 'namecheck' argument specifies the
118 * expected type (pool, dataset, or none).
121 * The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or
122 * NULL if no input nvlist was provided. Changes to this nvlist are
123 * ignored. If the input nvlist could not be deserialized, the
124 * ioctl will fail and the callback will not be called.
127 * The output nvlist, initially empty. The callback can fill it in,
128 * and it will be returned to userland by serializing it into
129 * zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization
130 * fails (e.g. because the caller didn't supply a large enough
131 * buffer), then the overall ioctl will fail. See the
132 * 'smush_nvlist' argument above for additional behaviors.
134 * There are two typical uses of the output nvlist:
135 * - To return state, e.g. property values. In this case,
136 * smush_outnvlist should be false. If the buffer was not large
137 * enough, the caller will reallocate a larger buffer and try
140 * - To return multiple errors from an ioctl which makes on-disk
141 * changes. In this case, smush_outnvlist should be true.
142 * Ioctls which make on-disk modifications should generally not
143 * use the outnvl if they succeed, because the caller can not
144 * distinguish between the operation failing, and
145 * deserialization failing.
147 * IOCTL Interface Errors
149 * The following ioctl input errors can be returned:
150 * ZFS_ERR_IOC_CMD_UNAVAIL the ioctl number is not supported by kernel
151 * ZFS_ERR_IOC_ARG_UNAVAIL an input argument is not supported by kernel
152 * ZFS_ERR_IOC_ARG_REQUIRED a required input argument is missing
153 * ZFS_ERR_IOC_ARG_BADTYPE an input argument has an invalid type
156 #include <sys/types.h>
157 #include <sys/param.h>
158 #include <sys/errno.h>
160 #include <sys/file.h>
161 #include <sys/kmem.h>
162 #include <sys/cmn_err.h>
163 #include <sys/stat.h>
164 #include <sys/zfs_ioctl.h>
165 #include <sys/zfs_vfsops.h>
166 #include <sys/zfs_znode.h>
169 #include <sys/spa_impl.h>
170 #include <sys/vdev.h>
171 #include <sys/vdev_impl.h>
173 #include <sys/dsl_dir.h>
174 #include <sys/dsl_dataset.h>
175 #include <sys/dsl_prop.h>
176 #include <sys/dsl_deleg.h>
177 #include <sys/dmu_objset.h>
178 #include <sys/dmu_impl.h>
179 #include <sys/dmu_redact.h>
180 #include <sys/dmu_tx.h>
181 #include <sys/sunddi.h>
182 #include <sys/policy.h>
183 #include <sys/zone.h>
184 #include <sys/nvpair.h>
185 #include <sys/pathname.h>
187 #include <sys/fs/zfs.h>
188 #include <sys/zfs_ctldir.h>
189 #include <sys/zfs_dir.h>
190 #include <sys/zfs_onexit.h>
191 #include <sys/zvol.h>
192 #include <sys/dsl_scan.h>
193 #include <sys/fm/util.h>
194 #include <sys/dsl_crypt.h>
195 #include <sys/rrwlock.h>
197 #include <sys/dmu_recv.h>
198 #include <sys/dmu_send.h>
199 #include <sys/dmu_recv.h>
200 #include <sys/dsl_destroy.h>
201 #include <sys/dsl_bookmark.h>
202 #include <sys/dsl_userhold.h>
203 #include <sys/zfeature.h>
205 #include <sys/zio_checksum.h>
206 #include <sys/vdev_removal.h>
207 #include <sys/vdev_impl.h>
208 #include <sys/vdev_initialize.h>
209 #include <sys/vdev_trim.h>
211 #include "zfs_namecheck.h"
212 #include "zfs_prop.h"
213 #include "zfs_deleg.h"
214 #include "zfs_comutil.h"
216 #include <sys/lua/lua.h>
217 #include <sys/lua/lauxlib.h>
218 #include <sys/zfs_ioctl_impl.h>
220 kmutex_t zfsdev_state_lock
;
221 zfsdev_state_t
*zfsdev_state_list
;
223 extern void zfs_init(void);
224 extern void zfs_fini(void);
227 * Limit maximum nvlist size. We don't want users passing in insane values
228 * for zc->zc_nvlist_src_size, since we will need to allocate that much memory.
230 #define MAX_NVLIST_SRC_SIZE KMALLOC_MAX_SIZE
232 uint_t zfs_fsyncer_key
;
233 uint_t zfs_allow_log_key
;
235 /* DATA_TYPE_ANY is used when zkey_type can vary. */
236 #define DATA_TYPE_ANY DATA_TYPE_UNKNOWN
238 typedef struct zfs_ioc_vec
{
239 zfs_ioc_legacy_func_t
*zvec_legacy_func
;
240 zfs_ioc_func_t
*zvec_func
;
241 zfs_secpolicy_func_t
*zvec_secpolicy
;
242 zfs_ioc_namecheck_t zvec_namecheck
;
243 boolean_t zvec_allow_log
;
244 zfs_ioc_poolcheck_t zvec_pool_check
;
245 boolean_t zvec_smush_outnvlist
;
246 const char *zvec_name
;
247 const zfs_ioc_key_t
*zvec_nvl_keys
;
248 size_t zvec_nvl_key_count
;
251 /* This array is indexed by zfs_userquota_prop_t */
252 static const char *userquota_perms
[] = {
253 ZFS_DELEG_PERM_USERUSED
,
254 ZFS_DELEG_PERM_USERQUOTA
,
255 ZFS_DELEG_PERM_GROUPUSED
,
256 ZFS_DELEG_PERM_GROUPQUOTA
,
257 ZFS_DELEG_PERM_USEROBJUSED
,
258 ZFS_DELEG_PERM_USEROBJQUOTA
,
259 ZFS_DELEG_PERM_GROUPOBJUSED
,
260 ZFS_DELEG_PERM_GROUPOBJQUOTA
,
261 ZFS_DELEG_PERM_PROJECTUSED
,
262 ZFS_DELEG_PERM_PROJECTQUOTA
,
263 ZFS_DELEG_PERM_PROJECTOBJUSED
,
264 ZFS_DELEG_PERM_PROJECTOBJQUOTA
,
267 static int zfs_ioc_userspace_upgrade(zfs_cmd_t
*zc
);
268 static int zfs_ioc_id_quota_upgrade(zfs_cmd_t
*zc
);
269 static int zfs_check_settable(const char *name
, nvpair_t
*property
,
271 static int zfs_check_clearable(char *dataset
, nvlist_t
*props
,
273 static int zfs_fill_zplprops_root(uint64_t, nvlist_t
*, nvlist_t
*,
275 int zfs_set_prop_nvlist(const char *, zprop_source_t
, nvlist_t
*, nvlist_t
*);
276 static int get_nvlist(uint64_t nvl
, uint64_t size
, int iflag
, nvlist_t
**nvp
);
279 history_str_free(char *buf
)
281 kmem_free(buf
, HIS_MAX_RECORD_LEN
);
285 history_str_get(zfs_cmd_t
*zc
)
289 if (zc
->zc_history
== 0)
292 buf
= kmem_alloc(HIS_MAX_RECORD_LEN
, KM_SLEEP
);
293 if (copyinstr((void *)(uintptr_t)zc
->zc_history
,
294 buf
, HIS_MAX_RECORD_LEN
, NULL
) != 0) {
295 history_str_free(buf
);
299 buf
[HIS_MAX_RECORD_LEN
-1] = '\0';
305 * Check to see if the named dataset is currently defined as bootable
308 zfs_is_bootfs(const char *name
)
312 if (dmu_objset_hold(name
, FTAG
, &os
) == 0) {
314 ret
= (dmu_objset_id(os
) == spa_bootfs(dmu_objset_spa(os
)));
315 dmu_objset_rele(os
, FTAG
);
322 * Return non-zero if the spa version is less than requested version.
325 zfs_earlier_version(const char *name
, int version
)
329 if (spa_open(name
, &spa
, FTAG
) == 0) {
330 if (spa_version(spa
) < version
) {
331 spa_close(spa
, FTAG
);
334 spa_close(spa
, FTAG
);
340 * Return TRUE if the ZPL version is less than requested version.
343 zpl_earlier_version(const char *name
, int version
)
346 boolean_t rc
= B_TRUE
;
348 if (dmu_objset_hold(name
, FTAG
, &os
) == 0) {
351 if (dmu_objset_type(os
) != DMU_OST_ZFS
) {
352 dmu_objset_rele(os
, FTAG
);
355 /* XXX reading from non-owned objset */
356 if (zfs_get_zplprop(os
, ZFS_PROP_VERSION
, &zplversion
) == 0)
357 rc
= zplversion
< version
;
358 dmu_objset_rele(os
, FTAG
);
364 zfs_log_history(zfs_cmd_t
*zc
)
369 if ((buf
= history_str_get(zc
)) == NULL
)
372 if (spa_open(zc
->zc_name
, &spa
, FTAG
) == 0) {
373 if (spa_version(spa
) >= SPA_VERSION_ZPOOL_HISTORY
)
374 (void) spa_history_log(spa
, buf
);
375 spa_close(spa
, FTAG
);
377 history_str_free(buf
);
381 * Policy for top-level read operations (list pools). Requires no privileges,
382 * and can be used in the local zone, as there is no associated dataset.
386 zfs_secpolicy_none(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
392 * Policy for dataset read operations (list children, get statistics). Requires
393 * no privileges, but must be visible in the local zone.
397 zfs_secpolicy_read(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
399 if (INGLOBALZONE(curproc
) ||
400 zone_dataset_visible(zc
->zc_name
, NULL
))
403 return (SET_ERROR(ENOENT
));
407 zfs_dozonecheck_impl(const char *dataset
, uint64_t zoned
, cred_t
*cr
)
412 * The dataset must be visible by this zone -- check this first
413 * so they don't see EPERM on something they shouldn't know about.
415 if (!INGLOBALZONE(curproc
) &&
416 !zone_dataset_visible(dataset
, &writable
))
417 return (SET_ERROR(ENOENT
));
419 if (INGLOBALZONE(curproc
)) {
421 * If the fs is zoned, only root can access it from the
424 if (secpolicy_zfs(cr
) && zoned
)
425 return (SET_ERROR(EPERM
));
428 * If we are in a local zone, the 'zoned' property must be set.
431 return (SET_ERROR(EPERM
));
433 /* must be writable by this zone */
435 return (SET_ERROR(EPERM
));
441 zfs_dozonecheck(const char *dataset
, cred_t
*cr
)
445 if (dsl_prop_get_integer(dataset
, zfs_prop_to_name(ZFS_PROP_ZONED
),
447 return (SET_ERROR(ENOENT
));
449 return (zfs_dozonecheck_impl(dataset
, zoned
, cr
));
453 zfs_dozonecheck_ds(const char *dataset
, dsl_dataset_t
*ds
, cred_t
*cr
)
457 if (dsl_prop_get_int_ds(ds
, zfs_prop_to_name(ZFS_PROP_ZONED
), &zoned
))
458 return (SET_ERROR(ENOENT
));
460 return (zfs_dozonecheck_impl(dataset
, zoned
, cr
));
464 zfs_secpolicy_write_perms_ds(const char *name
, dsl_dataset_t
*ds
,
465 const char *perm
, cred_t
*cr
)
469 error
= zfs_dozonecheck_ds(name
, ds
, cr
);
471 error
= secpolicy_zfs(cr
);
473 error
= dsl_deleg_access_impl(ds
, perm
, cr
);
479 zfs_secpolicy_write_perms(const char *name
, const char *perm
, cred_t
*cr
)
486 * First do a quick check for root in the global zone, which
487 * is allowed to do all write_perms. This ensures that zfs_ioc_*
488 * will get to handle nonexistent datasets.
490 if (INGLOBALZONE(curproc
) && secpolicy_zfs(cr
) == 0)
493 error
= dsl_pool_hold(name
, FTAG
, &dp
);
497 error
= dsl_dataset_hold(dp
, name
, FTAG
, &ds
);
499 dsl_pool_rele(dp
, FTAG
);
503 error
= zfs_secpolicy_write_perms_ds(name
, ds
, perm
, cr
);
505 dsl_dataset_rele(ds
, FTAG
);
506 dsl_pool_rele(dp
, FTAG
);
511 * Policy for setting the security label property.
513 * Returns 0 for success, non-zero for access and other errors.
516 zfs_set_slabel_policy(const char *name
, char *strval
, cred_t
*cr
)
519 char ds_hexsl
[MAXNAMELEN
];
520 bslabel_t ds_sl
, new_sl
;
521 boolean_t new_default
= FALSE
;
523 int needed_priv
= -1;
526 /* First get the existing dataset label. */
527 error
= dsl_prop_get(name
, zfs_prop_to_name(ZFS_PROP_MLSLABEL
),
528 1, sizeof (ds_hexsl
), &ds_hexsl
, NULL
);
530 return (SET_ERROR(EPERM
));
532 if (strcasecmp(strval
, ZFS_MLSLABEL_DEFAULT
) == 0)
535 /* The label must be translatable */
536 if (!new_default
&& (hexstr_to_label(strval
, &new_sl
) != 0))
537 return (SET_ERROR(EINVAL
));
540 * In a non-global zone, disallow attempts to set a label that
541 * doesn't match that of the zone; otherwise no other checks
544 if (!INGLOBALZONE(curproc
)) {
545 if (new_default
|| !blequal(&new_sl
, CR_SL(CRED())))
546 return (SET_ERROR(EPERM
));
551 * For global-zone datasets (i.e., those whose zoned property is
552 * "off", verify that the specified new label is valid for the
555 if (dsl_prop_get_integer(name
,
556 zfs_prop_to_name(ZFS_PROP_ZONED
), &zoned
, NULL
))
557 return (SET_ERROR(EPERM
));
559 if (zfs_check_global_label(name
, strval
) != 0)
560 return (SET_ERROR(EPERM
));
564 * If the existing dataset label is nondefault, check if the
565 * dataset is mounted (label cannot be changed while mounted).
566 * Get the zfsvfs_t; if there isn't one, then the dataset isn't
567 * mounted (or isn't a dataset, doesn't exist, ...).
569 if (strcasecmp(ds_hexsl
, ZFS_MLSLABEL_DEFAULT
) != 0) {
571 static char *setsl_tag
= "setsl_tag";
574 * Try to own the dataset; abort if there is any error,
575 * (e.g., already mounted, in use, or other error).
577 error
= dmu_objset_own(name
, DMU_OST_ZFS
, B_TRUE
, B_TRUE
,
580 return (SET_ERROR(EPERM
));
582 dmu_objset_disown(os
, B_TRUE
, setsl_tag
);
585 needed_priv
= PRIV_FILE_DOWNGRADE_SL
;
589 if (hexstr_to_label(strval
, &new_sl
) != 0)
590 return (SET_ERROR(EPERM
));
592 if (blstrictdom(&ds_sl
, &new_sl
))
593 needed_priv
= PRIV_FILE_DOWNGRADE_SL
;
594 else if (blstrictdom(&new_sl
, &ds_sl
))
595 needed_priv
= PRIV_FILE_UPGRADE_SL
;
597 /* dataset currently has a default label */
599 needed_priv
= PRIV_FILE_UPGRADE_SL
;
603 if (needed_priv
!= -1)
604 return (PRIV_POLICY(cr
, needed_priv
, B_FALSE
, EPERM
, NULL
));
607 return (SET_ERROR(ENOTSUP
));
608 #endif /* HAVE_MLSLABEL */
612 zfs_secpolicy_setprop(const char *dsname
, zfs_prop_t prop
, nvpair_t
*propval
,
618 * Check permissions for special properties.
625 * Disallow setting of 'zoned' from within a local zone.
627 if (!INGLOBALZONE(curproc
))
628 return (SET_ERROR(EPERM
));
632 case ZFS_PROP_FILESYSTEM_LIMIT
:
633 case ZFS_PROP_SNAPSHOT_LIMIT
:
634 if (!INGLOBALZONE(curproc
)) {
636 char setpoint
[ZFS_MAX_DATASET_NAME_LEN
];
638 * Unprivileged users are allowed to modify the
639 * limit on things *under* (ie. contained by)
640 * the thing they own.
642 if (dsl_prop_get_integer(dsname
,
643 zfs_prop_to_name(ZFS_PROP_ZONED
), &zoned
, setpoint
))
644 return (SET_ERROR(EPERM
));
645 if (!zoned
|| strlen(dsname
) <= strlen(setpoint
))
646 return (SET_ERROR(EPERM
));
650 case ZFS_PROP_MLSLABEL
:
651 if (!is_system_labeled())
652 return (SET_ERROR(EPERM
));
654 if (nvpair_value_string(propval
, &strval
) == 0) {
657 err
= zfs_set_slabel_policy(dsname
, strval
, CRED());
664 return (zfs_secpolicy_write_perms(dsname
, zfs_prop_to_name(prop
), cr
));
669 zfs_secpolicy_set_fsacl(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
673 error
= zfs_dozonecheck(zc
->zc_name
, cr
);
678 * permission to set permissions will be evaluated later in
679 * dsl_deleg_can_allow()
686 zfs_secpolicy_rollback(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
688 return (zfs_secpolicy_write_perms(zc
->zc_name
,
689 ZFS_DELEG_PERM_ROLLBACK
, cr
));
694 zfs_secpolicy_send(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
702 * Generate the current snapshot name from the given objsetid, then
703 * use that name for the secpolicy/zone checks.
705 cp
= strchr(zc
->zc_name
, '@');
707 return (SET_ERROR(EINVAL
));
708 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
712 error
= dsl_dataset_hold_obj(dp
, zc
->zc_sendobj
, FTAG
, &ds
);
714 dsl_pool_rele(dp
, FTAG
);
718 dsl_dataset_name(ds
, zc
->zc_name
);
720 error
= zfs_secpolicy_write_perms_ds(zc
->zc_name
, ds
,
721 ZFS_DELEG_PERM_SEND
, cr
);
722 dsl_dataset_rele(ds
, FTAG
);
723 dsl_pool_rele(dp
, FTAG
);
730 zfs_secpolicy_send_new(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
732 return (zfs_secpolicy_write_perms(zc
->zc_name
,
733 ZFS_DELEG_PERM_SEND
, cr
));
737 zfs_secpolicy_share(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
739 return (SET_ERROR(ENOTSUP
));
743 zfs_secpolicy_smb_acl(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
745 return (SET_ERROR(ENOTSUP
));
749 zfs_get_parent(const char *datasetname
, char *parent
, int parentsize
)
754 * Remove the @bla or /bla from the end of the name to get the parent.
756 (void) strncpy(parent
, datasetname
, parentsize
);
757 cp
= strrchr(parent
, '@');
761 cp
= strrchr(parent
, '/');
763 return (SET_ERROR(ENOENT
));
771 zfs_secpolicy_destroy_perms(const char *name
, cred_t
*cr
)
775 if ((error
= zfs_secpolicy_write_perms(name
,
776 ZFS_DELEG_PERM_MOUNT
, cr
)) != 0)
779 return (zfs_secpolicy_write_perms(name
, ZFS_DELEG_PERM_DESTROY
, cr
));
784 zfs_secpolicy_destroy(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
786 return (zfs_secpolicy_destroy_perms(zc
->zc_name
, cr
));
790 * Destroying snapshots with delegated permissions requires
791 * descendant mount and destroy permissions.
795 zfs_secpolicy_destroy_snaps(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
798 nvpair_t
*pair
, *nextpair
;
801 snaps
= fnvlist_lookup_nvlist(innvl
, "snaps");
803 for (pair
= nvlist_next_nvpair(snaps
, NULL
); pair
!= NULL
;
805 nextpair
= nvlist_next_nvpair(snaps
, pair
);
806 error
= zfs_secpolicy_destroy_perms(nvpair_name(pair
), cr
);
807 if (error
== ENOENT
) {
809 * Ignore any snapshots that don't exist (we consider
810 * them "already destroyed"). Remove the name from the
811 * nvl here in case the snapshot is created between
812 * now and when we try to destroy it (in which case
813 * we don't want to destroy it since we haven't
814 * checked for permission).
816 fnvlist_remove_nvpair(snaps
, pair
);
827 zfs_secpolicy_rename_perms(const char *from
, const char *to
, cred_t
*cr
)
829 char parentname
[ZFS_MAX_DATASET_NAME_LEN
];
832 if ((error
= zfs_secpolicy_write_perms(from
,
833 ZFS_DELEG_PERM_RENAME
, cr
)) != 0)
836 if ((error
= zfs_secpolicy_write_perms(from
,
837 ZFS_DELEG_PERM_MOUNT
, cr
)) != 0)
840 if ((error
= zfs_get_parent(to
, parentname
,
841 sizeof (parentname
))) != 0)
844 if ((error
= zfs_secpolicy_write_perms(parentname
,
845 ZFS_DELEG_PERM_CREATE
, cr
)) != 0)
848 if ((error
= zfs_secpolicy_write_perms(parentname
,
849 ZFS_DELEG_PERM_MOUNT
, cr
)) != 0)
857 zfs_secpolicy_rename(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
859 return (zfs_secpolicy_rename_perms(zc
->zc_name
, zc
->zc_value
, cr
));
864 zfs_secpolicy_promote(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
867 dsl_dataset_t
*clone
;
870 error
= zfs_secpolicy_write_perms(zc
->zc_name
,
871 ZFS_DELEG_PERM_PROMOTE
, cr
);
875 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
879 error
= dsl_dataset_hold(dp
, zc
->zc_name
, FTAG
, &clone
);
882 char parentname
[ZFS_MAX_DATASET_NAME_LEN
];
883 dsl_dataset_t
*origin
= NULL
;
887 error
= dsl_dataset_hold_obj(dd
->dd_pool
,
888 dsl_dir_phys(dd
)->dd_origin_obj
, FTAG
, &origin
);
890 dsl_dataset_rele(clone
, FTAG
);
891 dsl_pool_rele(dp
, FTAG
);
895 error
= zfs_secpolicy_write_perms_ds(zc
->zc_name
, clone
,
896 ZFS_DELEG_PERM_MOUNT
, cr
);
898 dsl_dataset_name(origin
, parentname
);
900 error
= zfs_secpolicy_write_perms_ds(parentname
, origin
,
901 ZFS_DELEG_PERM_PROMOTE
, cr
);
903 dsl_dataset_rele(clone
, FTAG
);
904 dsl_dataset_rele(origin
, FTAG
);
906 dsl_pool_rele(dp
, FTAG
);
912 zfs_secpolicy_recv(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
916 if ((error
= zfs_secpolicy_write_perms(zc
->zc_name
,
917 ZFS_DELEG_PERM_RECEIVE
, cr
)) != 0)
920 if ((error
= zfs_secpolicy_write_perms(zc
->zc_name
,
921 ZFS_DELEG_PERM_MOUNT
, cr
)) != 0)
924 return (zfs_secpolicy_write_perms(zc
->zc_name
,
925 ZFS_DELEG_PERM_CREATE
, cr
));
930 zfs_secpolicy_recv_new(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
932 return (zfs_secpolicy_recv(zc
, innvl
, cr
));
936 zfs_secpolicy_snapshot_perms(const char *name
, cred_t
*cr
)
938 return (zfs_secpolicy_write_perms(name
,
939 ZFS_DELEG_PERM_SNAPSHOT
, cr
));
943 * Check for permission to create each snapshot in the nvlist.
947 zfs_secpolicy_snapshot(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
953 snaps
= fnvlist_lookup_nvlist(innvl
, "snaps");
955 for (pair
= nvlist_next_nvpair(snaps
, NULL
); pair
!= NULL
;
956 pair
= nvlist_next_nvpair(snaps
, pair
)) {
957 char *name
= nvpair_name(pair
);
958 char *atp
= strchr(name
, '@');
961 error
= SET_ERROR(EINVAL
);
965 error
= zfs_secpolicy_snapshot_perms(name
, cr
);
974 * Check for permission to create each bookmark in the nvlist.
978 zfs_secpolicy_bookmark(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
982 for (nvpair_t
*pair
= nvlist_next_nvpair(innvl
, NULL
);
983 pair
!= NULL
; pair
= nvlist_next_nvpair(innvl
, pair
)) {
984 char *name
= nvpair_name(pair
);
985 char *hashp
= strchr(name
, '#');
988 error
= SET_ERROR(EINVAL
);
992 error
= zfs_secpolicy_write_perms(name
,
993 ZFS_DELEG_PERM_BOOKMARK
, cr
);
1003 zfs_secpolicy_destroy_bookmarks(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1005 nvpair_t
*pair
, *nextpair
;
1008 for (pair
= nvlist_next_nvpair(innvl
, NULL
); pair
!= NULL
;
1010 char *name
= nvpair_name(pair
);
1011 char *hashp
= strchr(name
, '#');
1012 nextpair
= nvlist_next_nvpair(innvl
, pair
);
1014 if (hashp
== NULL
) {
1015 error
= SET_ERROR(EINVAL
);
1020 error
= zfs_secpolicy_write_perms(name
,
1021 ZFS_DELEG_PERM_DESTROY
, cr
);
1023 if (error
== ENOENT
) {
1025 * Ignore any filesystems that don't exist (we consider
1026 * their bookmarks "already destroyed"). Remove
1027 * the name from the nvl here in case the filesystem
1028 * is created between now and when we try to destroy
1029 * the bookmark (in which case we don't want to
1030 * destroy it since we haven't checked for permission).
1032 fnvlist_remove_nvpair(innvl
, pair
);
1044 zfs_secpolicy_log_history(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1047 * Even root must have a proper TSD so that we know what pool
1050 if (tsd_get(zfs_allow_log_key
) == NULL
)
1051 return (SET_ERROR(EPERM
));
1056 zfs_secpolicy_create_clone(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1058 char parentname
[ZFS_MAX_DATASET_NAME_LEN
];
1062 if ((error
= zfs_get_parent(zc
->zc_name
, parentname
,
1063 sizeof (parentname
))) != 0)
1066 if (nvlist_lookup_string(innvl
, "origin", &origin
) == 0 &&
1067 (error
= zfs_secpolicy_write_perms(origin
,
1068 ZFS_DELEG_PERM_CLONE
, cr
)) != 0)
1071 if ((error
= zfs_secpolicy_write_perms(parentname
,
1072 ZFS_DELEG_PERM_CREATE
, cr
)) != 0)
1075 return (zfs_secpolicy_write_perms(parentname
,
1076 ZFS_DELEG_PERM_MOUNT
, cr
));
1080 * Policy for pool operations - create/destroy pools, add vdevs, etc. Requires
1081 * SYS_CONFIG privilege, which is not available in a local zone.
1085 zfs_secpolicy_config(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1087 if (secpolicy_sys_config(cr
, B_FALSE
) != 0)
1088 return (SET_ERROR(EPERM
));
1094 * Policy for object to name lookups.
1098 zfs_secpolicy_diff(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1102 if ((error
= secpolicy_sys_config(cr
, B_FALSE
)) == 0)
1105 error
= zfs_secpolicy_write_perms(zc
->zc_name
, ZFS_DELEG_PERM_DIFF
, cr
);
1110 * Policy for fault injection. Requires all privileges.
1114 zfs_secpolicy_inject(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1116 return (secpolicy_zinject(cr
));
1121 zfs_secpolicy_inherit_prop(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1123 zfs_prop_t prop
= zfs_name_to_prop(zc
->zc_value
);
1125 if (prop
== ZPROP_INVAL
) {
1126 if (!zfs_prop_user(zc
->zc_value
))
1127 return (SET_ERROR(EINVAL
));
1128 return (zfs_secpolicy_write_perms(zc
->zc_name
,
1129 ZFS_DELEG_PERM_USERPROP
, cr
));
1131 return (zfs_secpolicy_setprop(zc
->zc_name
, prop
,
1137 zfs_secpolicy_userspace_one(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1139 int err
= zfs_secpolicy_read(zc
, innvl
, cr
);
1143 if (zc
->zc_objset_type
>= ZFS_NUM_USERQUOTA_PROPS
)
1144 return (SET_ERROR(EINVAL
));
1146 if (zc
->zc_value
[0] == 0) {
1148 * They are asking about a posix uid/gid. If it's
1149 * themself, allow it.
1151 if (zc
->zc_objset_type
== ZFS_PROP_USERUSED
||
1152 zc
->zc_objset_type
== ZFS_PROP_USERQUOTA
||
1153 zc
->zc_objset_type
== ZFS_PROP_USEROBJUSED
||
1154 zc
->zc_objset_type
== ZFS_PROP_USEROBJQUOTA
) {
1155 if (zc
->zc_guid
== crgetuid(cr
))
1157 } else if (zc
->zc_objset_type
== ZFS_PROP_GROUPUSED
||
1158 zc
->zc_objset_type
== ZFS_PROP_GROUPQUOTA
||
1159 zc
->zc_objset_type
== ZFS_PROP_GROUPOBJUSED
||
1160 zc
->zc_objset_type
== ZFS_PROP_GROUPOBJQUOTA
) {
1161 if (groupmember(zc
->zc_guid
, cr
))
1164 /* else is for project quota/used */
1167 return (zfs_secpolicy_write_perms(zc
->zc_name
,
1168 userquota_perms
[zc
->zc_objset_type
], cr
));
1172 zfs_secpolicy_userspace_many(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1174 int err
= zfs_secpolicy_read(zc
, innvl
, cr
);
1178 if (zc
->zc_objset_type
>= ZFS_NUM_USERQUOTA_PROPS
)
1179 return (SET_ERROR(EINVAL
));
1181 return (zfs_secpolicy_write_perms(zc
->zc_name
,
1182 userquota_perms
[zc
->zc_objset_type
], cr
));
1187 zfs_secpolicy_userspace_upgrade(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1189 return (zfs_secpolicy_setprop(zc
->zc_name
, ZFS_PROP_VERSION
,
1195 zfs_secpolicy_hold(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1201 holds
= fnvlist_lookup_nvlist(innvl
, "holds");
1203 for (pair
= nvlist_next_nvpair(holds
, NULL
); pair
!= NULL
;
1204 pair
= nvlist_next_nvpair(holds
, pair
)) {
1205 char fsname
[ZFS_MAX_DATASET_NAME_LEN
];
1206 error
= dmu_fsname(nvpair_name(pair
), fsname
);
1209 error
= zfs_secpolicy_write_perms(fsname
,
1210 ZFS_DELEG_PERM_HOLD
, cr
);
1219 zfs_secpolicy_release(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1224 for (pair
= nvlist_next_nvpair(innvl
, NULL
); pair
!= NULL
;
1225 pair
= nvlist_next_nvpair(innvl
, pair
)) {
1226 char fsname
[ZFS_MAX_DATASET_NAME_LEN
];
1227 error
= dmu_fsname(nvpair_name(pair
), fsname
);
1230 error
= zfs_secpolicy_write_perms(fsname
,
1231 ZFS_DELEG_PERM_RELEASE
, cr
);
1239 * Policy for allowing temporary snapshots to be taken or released
1242 zfs_secpolicy_tmp_snapshot(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1245 * A temporary snapshot is the same as a snapshot,
1246 * hold, destroy and release all rolled into one.
1247 * Delegated diff alone is sufficient that we allow this.
1251 if ((error
= zfs_secpolicy_write_perms(zc
->zc_name
,
1252 ZFS_DELEG_PERM_DIFF
, cr
)) == 0)
1255 error
= zfs_secpolicy_snapshot_perms(zc
->zc_name
, cr
);
1257 if (innvl
!= NULL
) {
1259 error
= zfs_secpolicy_hold(zc
, innvl
, cr
);
1261 error
= zfs_secpolicy_release(zc
, innvl
, cr
);
1263 error
= zfs_secpolicy_destroy(zc
, innvl
, cr
);
1269 zfs_secpolicy_load_key(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1271 return (zfs_secpolicy_write_perms(zc
->zc_name
,
1272 ZFS_DELEG_PERM_LOAD_KEY
, cr
));
1276 zfs_secpolicy_change_key(zfs_cmd_t
*zc
, nvlist_t
*innvl
, cred_t
*cr
)
1278 return (zfs_secpolicy_write_perms(zc
->zc_name
,
1279 ZFS_DELEG_PERM_CHANGE_KEY
, cr
));
1283 * Returns the nvlist as specified by the user in the zfs_cmd_t.
1286 get_nvlist(uint64_t nvl
, uint64_t size
, int iflag
, nvlist_t
**nvp
)
1290 nvlist_t
*list
= NULL
;
1293 * Read in and unpack the user-supplied nvlist.
1296 return (SET_ERROR(EINVAL
));
1298 packed
= vmem_alloc(size
, KM_SLEEP
);
1300 if ((error
= ddi_copyin((void *)(uintptr_t)nvl
, packed
, size
,
1302 vmem_free(packed
, size
);
1303 return (SET_ERROR(EFAULT
));
1306 if ((error
= nvlist_unpack(packed
, size
, &list
, 0)) != 0) {
1307 vmem_free(packed
, size
);
1311 vmem_free(packed
, size
);
1318 * Reduce the size of this nvlist until it can be serialized in 'max' bytes.
1319 * Entries will be removed from the end of the nvlist, and one int32 entry
1320 * named "N_MORE_ERRORS" will be added indicating how many entries were
1324 nvlist_smush(nvlist_t
*errors
, size_t max
)
1328 size
= fnvlist_size(errors
);
1331 nvpair_t
*more_errors
;
1335 return (SET_ERROR(ENOMEM
));
1337 fnvlist_add_int32(errors
, ZPROP_N_MORE_ERRORS
, 0);
1338 more_errors
= nvlist_prev_nvpair(errors
, NULL
);
1341 nvpair_t
*pair
= nvlist_prev_nvpair(errors
,
1343 fnvlist_remove_nvpair(errors
, pair
);
1345 size
= fnvlist_size(errors
);
1346 } while (size
> max
);
1348 fnvlist_remove_nvpair(errors
, more_errors
);
1349 fnvlist_add_int32(errors
, ZPROP_N_MORE_ERRORS
, n
);
1350 ASSERT3U(fnvlist_size(errors
), <=, max
);
1357 put_nvlist(zfs_cmd_t
*zc
, nvlist_t
*nvl
)
1359 char *packed
= NULL
;
1363 size
= fnvlist_size(nvl
);
1365 if (size
> zc
->zc_nvlist_dst_size
) {
1366 error
= SET_ERROR(ENOMEM
);
1368 packed
= fnvlist_pack(nvl
, &size
);
1369 if (ddi_copyout(packed
, (void *)(uintptr_t)zc
->zc_nvlist_dst
,
1370 size
, zc
->zc_iflags
) != 0)
1371 error
= SET_ERROR(EFAULT
);
1372 fnvlist_pack_free(packed
, size
);
1375 zc
->zc_nvlist_dst_size
= size
;
1376 zc
->zc_nvlist_dst_filled
= B_TRUE
;
1381 getzfsvfs_impl(objset_t
*os
, zfsvfs_t
**zfvp
)
1384 if (dmu_objset_type(os
) != DMU_OST_ZFS
) {
1385 return (SET_ERROR(EINVAL
));
1388 mutex_enter(&os
->os_user_ptr_lock
);
1389 *zfvp
= dmu_objset_get_user(os
);
1390 /* bump s_active only when non-zero to prevent umount race */
1391 error
= zfs_vfs_ref(zfvp
);
1392 mutex_exit(&os
->os_user_ptr_lock
);
1397 getzfsvfs(const char *dsname
, zfsvfs_t
**zfvp
)
1402 error
= dmu_objset_hold(dsname
, FTAG
, &os
);
1406 error
= getzfsvfs_impl(os
, zfvp
);
1407 dmu_objset_rele(os
, FTAG
);
1412 * Find a zfsvfs_t for a mounted filesystem, or create our own, in which
1413 * case its z_sb will be NULL, and it will be opened as the owner.
1414 * If 'writer' is set, the z_teardown_lock will be held for RW_WRITER,
1415 * which prevents all inode ops from running.
1418 zfsvfs_hold(const char *name
, void *tag
, zfsvfs_t
**zfvp
, boolean_t writer
)
1422 if (getzfsvfs(name
, zfvp
) != 0)
1423 error
= zfsvfs_create(name
, B_FALSE
, zfvp
);
1425 rrm_enter(&(*zfvp
)->z_teardown_lock
, (writer
) ? RW_WRITER
:
1427 if ((*zfvp
)->z_unmounted
) {
1429 * XXX we could probably try again, since the unmounting
1430 * thread should be just about to disassociate the
1431 * objset from the zfsvfs.
1433 rrm_exit(&(*zfvp
)->z_teardown_lock
, tag
);
1434 return (SET_ERROR(EBUSY
));
1441 zfsvfs_rele(zfsvfs_t
*zfsvfs
, void *tag
)
1443 rrm_exit(&zfsvfs
->z_teardown_lock
, tag
);
1446 deactivate_super(zfsvfs
->z_sb
);
1448 dmu_objset_disown(zfsvfs
->z_os
, B_TRUE
, zfsvfs
);
1449 zfsvfs_free(zfsvfs
);
1454 zfs_ioc_pool_create(zfs_cmd_t
*zc
)
1457 nvlist_t
*config
, *props
= NULL
;
1458 nvlist_t
*rootprops
= NULL
;
1459 nvlist_t
*zplprops
= NULL
;
1460 dsl_crypto_params_t
*dcp
= NULL
;
1461 char *spa_name
= zc
->zc_name
;
1462 boolean_t unload_wkey
= B_TRUE
;
1464 if ((error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1465 zc
->zc_iflags
, &config
)))
1468 if (zc
->zc_nvlist_src_size
!= 0 && (error
=
1469 get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
1470 zc
->zc_iflags
, &props
))) {
1471 nvlist_free(config
);
1476 nvlist_t
*nvl
= NULL
;
1477 nvlist_t
*hidden_args
= NULL
;
1478 uint64_t version
= SPA_VERSION
;
1481 (void) nvlist_lookup_uint64(props
,
1482 zpool_prop_to_name(ZPOOL_PROP_VERSION
), &version
);
1483 if (!SPA_VERSION_IS_SUPPORTED(version
)) {
1484 error
= SET_ERROR(EINVAL
);
1485 goto pool_props_bad
;
1487 (void) nvlist_lookup_nvlist(props
, ZPOOL_ROOTFS_PROPS
, &nvl
);
1489 error
= nvlist_dup(nvl
, &rootprops
, KM_SLEEP
);
1491 goto pool_props_bad
;
1492 (void) nvlist_remove_all(props
, ZPOOL_ROOTFS_PROPS
);
1495 (void) nvlist_lookup_nvlist(props
, ZPOOL_HIDDEN_ARGS
,
1497 error
= dsl_crypto_params_create_nvlist(DCP_CMD_NONE
,
1498 rootprops
, hidden_args
, &dcp
);
1500 goto pool_props_bad
;
1501 (void) nvlist_remove_all(props
, ZPOOL_HIDDEN_ARGS
);
1503 VERIFY(nvlist_alloc(&zplprops
, NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
1504 error
= zfs_fill_zplprops_root(version
, rootprops
,
1507 goto pool_props_bad
;
1509 if (nvlist_lookup_string(props
,
1510 zpool_prop_to_name(ZPOOL_PROP_TNAME
), &tname
) == 0)
1514 error
= spa_create(zc
->zc_name
, config
, props
, zplprops
, dcp
);
1517 * Set the remaining root properties
1519 if (!error
&& (error
= zfs_set_prop_nvlist(spa_name
,
1520 ZPROP_SRC_LOCAL
, rootprops
, NULL
)) != 0) {
1521 (void) spa_destroy(spa_name
);
1522 unload_wkey
= B_FALSE
; /* spa_destroy() unloads wrapping keys */
1526 nvlist_free(rootprops
);
1527 nvlist_free(zplprops
);
1528 nvlist_free(config
);
1530 dsl_crypto_params_free(dcp
, unload_wkey
&& !!error
);
1536 zfs_ioc_pool_destroy(zfs_cmd_t
*zc
)
1539 zfs_log_history(zc
);
1540 error
= spa_destroy(zc
->zc_name
);
1546 zfs_ioc_pool_import(zfs_cmd_t
*zc
)
1548 nvlist_t
*config
, *props
= NULL
;
1552 if ((error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1553 zc
->zc_iflags
, &config
)) != 0)
1556 if (zc
->zc_nvlist_src_size
!= 0 && (error
=
1557 get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
1558 zc
->zc_iflags
, &props
))) {
1559 nvlist_free(config
);
1563 if (nvlist_lookup_uint64(config
, ZPOOL_CONFIG_POOL_GUID
, &guid
) != 0 ||
1564 guid
!= zc
->zc_guid
)
1565 error
= SET_ERROR(EINVAL
);
1567 error
= spa_import(zc
->zc_name
, config
, props
, zc
->zc_cookie
);
1569 if (zc
->zc_nvlist_dst
!= 0) {
1572 if ((err
= put_nvlist(zc
, config
)) != 0)
1576 nvlist_free(config
);
1583 zfs_ioc_pool_export(zfs_cmd_t
*zc
)
1586 boolean_t force
= (boolean_t
)zc
->zc_cookie
;
1587 boolean_t hardforce
= (boolean_t
)zc
->zc_guid
;
1589 zfs_log_history(zc
);
1590 error
= spa_export(zc
->zc_name
, NULL
, force
, hardforce
);
1596 zfs_ioc_pool_configs(zfs_cmd_t
*zc
)
1601 if ((configs
= spa_all_configs(&zc
->zc_cookie
)) == NULL
)
1602 return (SET_ERROR(EEXIST
));
1604 error
= put_nvlist(zc
, configs
);
1606 nvlist_free(configs
);
1613 * zc_name name of the pool
1616 * zc_cookie real errno
1617 * zc_nvlist_dst config nvlist
1618 * zc_nvlist_dst_size size of config nvlist
1621 zfs_ioc_pool_stats(zfs_cmd_t
*zc
)
1627 error
= spa_get_stats(zc
->zc_name
, &config
, zc
->zc_value
,
1628 sizeof (zc
->zc_value
));
1630 if (config
!= NULL
) {
1631 ret
= put_nvlist(zc
, config
);
1632 nvlist_free(config
);
1635 * The config may be present even if 'error' is non-zero.
1636 * In this case we return success, and preserve the real errno
1639 zc
->zc_cookie
= error
;
1648 * Try to import the given pool, returning pool stats as appropriate so that
1649 * user land knows which devices are available and overall pool health.
1652 zfs_ioc_pool_tryimport(zfs_cmd_t
*zc
)
1654 nvlist_t
*tryconfig
, *config
= NULL
;
1657 if ((error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1658 zc
->zc_iflags
, &tryconfig
)) != 0)
1661 config
= spa_tryimport(tryconfig
);
1663 nvlist_free(tryconfig
);
1666 return (SET_ERROR(EINVAL
));
1668 error
= put_nvlist(zc
, config
);
1669 nvlist_free(config
);
1676 * zc_name name of the pool
1677 * zc_cookie scan func (pool_scan_func_t)
1678 * zc_flags scrub pause/resume flag (pool_scrub_cmd_t)
1681 zfs_ioc_pool_scan(zfs_cmd_t
*zc
)
1686 if (zc
->zc_flags
>= POOL_SCRUB_FLAGS_END
)
1687 return (SET_ERROR(EINVAL
));
1689 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1692 if (zc
->zc_flags
== POOL_SCRUB_PAUSE
)
1693 error
= spa_scrub_pause_resume(spa
, POOL_SCRUB_PAUSE
);
1694 else if (zc
->zc_cookie
== POOL_SCAN_NONE
)
1695 error
= spa_scan_stop(spa
);
1697 error
= spa_scan(spa
, zc
->zc_cookie
);
1699 spa_close(spa
, FTAG
);
1705 zfs_ioc_pool_freeze(zfs_cmd_t
*zc
)
1710 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
1713 spa_close(spa
, FTAG
);
1719 zfs_ioc_pool_upgrade(zfs_cmd_t
*zc
)
1724 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1727 if (zc
->zc_cookie
< spa_version(spa
) ||
1728 !SPA_VERSION_IS_SUPPORTED(zc
->zc_cookie
)) {
1729 spa_close(spa
, FTAG
);
1730 return (SET_ERROR(EINVAL
));
1733 spa_upgrade(spa
, zc
->zc_cookie
);
1734 spa_close(spa
, FTAG
);
1740 zfs_ioc_pool_get_history(zfs_cmd_t
*zc
)
1747 if ((size
= zc
->zc_history_len
) == 0)
1748 return (SET_ERROR(EINVAL
));
1750 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1753 if (spa_version(spa
) < SPA_VERSION_ZPOOL_HISTORY
) {
1754 spa_close(spa
, FTAG
);
1755 return (SET_ERROR(ENOTSUP
));
1758 hist_buf
= vmem_alloc(size
, KM_SLEEP
);
1759 if ((error
= spa_history_get(spa
, &zc
->zc_history_offset
,
1760 &zc
->zc_history_len
, hist_buf
)) == 0) {
1761 error
= ddi_copyout(hist_buf
,
1762 (void *)(uintptr_t)zc
->zc_history
,
1763 zc
->zc_history_len
, zc
->zc_iflags
);
1766 spa_close(spa
, FTAG
);
1767 vmem_free(hist_buf
, size
);
1772 zfs_ioc_pool_reguid(zfs_cmd_t
*zc
)
1777 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
1779 error
= spa_change_guid(spa
);
1780 spa_close(spa
, FTAG
);
1786 zfs_ioc_dsobj_to_dsname(zfs_cmd_t
*zc
)
1788 return (dsl_dsobj_to_dsname(zc
->zc_name
, zc
->zc_obj
, zc
->zc_value
));
1793 * zc_name name of filesystem
1794 * zc_obj object to find
1797 * zc_value name of object
1800 zfs_ioc_obj_to_path(zfs_cmd_t
*zc
)
1805 /* XXX reading from objset not owned */
1806 if ((error
= dmu_objset_hold_flags(zc
->zc_name
, B_TRUE
,
1809 if (dmu_objset_type(os
) != DMU_OST_ZFS
) {
1810 dmu_objset_rele_flags(os
, B_TRUE
, FTAG
);
1811 return (SET_ERROR(EINVAL
));
1813 error
= zfs_obj_to_path(os
, zc
->zc_obj
, zc
->zc_value
,
1814 sizeof (zc
->zc_value
));
1815 dmu_objset_rele_flags(os
, B_TRUE
, FTAG
);
1822 * zc_name name of filesystem
1823 * zc_obj object to find
1826 * zc_stat stats on object
1827 * zc_value path to object
1830 zfs_ioc_obj_to_stats(zfs_cmd_t
*zc
)
1835 /* XXX reading from objset not owned */
1836 if ((error
= dmu_objset_hold_flags(zc
->zc_name
, B_TRUE
,
1839 if (dmu_objset_type(os
) != DMU_OST_ZFS
) {
1840 dmu_objset_rele_flags(os
, B_TRUE
, FTAG
);
1841 return (SET_ERROR(EINVAL
));
1843 error
= zfs_obj_to_stats(os
, zc
->zc_obj
, &zc
->zc_stat
, zc
->zc_value
,
1844 sizeof (zc
->zc_value
));
1845 dmu_objset_rele_flags(os
, B_TRUE
, FTAG
);
1851 zfs_ioc_vdev_add(zfs_cmd_t
*zc
)
1857 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
1861 error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1862 zc
->zc_iflags
, &config
);
1864 error
= spa_vdev_add(spa
, config
);
1865 nvlist_free(config
);
1867 spa_close(spa
, FTAG
);
1873 * zc_name name of the pool
1874 * zc_guid guid of vdev to remove
1875 * zc_cookie cancel removal
1878 zfs_ioc_vdev_remove(zfs_cmd_t
*zc
)
1883 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
1886 if (zc
->zc_cookie
!= 0) {
1887 error
= spa_vdev_remove_cancel(spa
);
1889 error
= spa_vdev_remove(spa
, zc
->zc_guid
, B_FALSE
);
1891 spa_close(spa
, FTAG
);
1896 zfs_ioc_vdev_set_state(zfs_cmd_t
*zc
)
1900 vdev_state_t newstate
= VDEV_STATE_UNKNOWN
;
1902 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1904 switch (zc
->zc_cookie
) {
1905 case VDEV_STATE_ONLINE
:
1906 error
= vdev_online(spa
, zc
->zc_guid
, zc
->zc_obj
, &newstate
);
1909 case VDEV_STATE_OFFLINE
:
1910 error
= vdev_offline(spa
, zc
->zc_guid
, zc
->zc_obj
);
1913 case VDEV_STATE_FAULTED
:
1914 if (zc
->zc_obj
!= VDEV_AUX_ERR_EXCEEDED
&&
1915 zc
->zc_obj
!= VDEV_AUX_EXTERNAL
&&
1916 zc
->zc_obj
!= VDEV_AUX_EXTERNAL_PERSIST
)
1917 zc
->zc_obj
= VDEV_AUX_ERR_EXCEEDED
;
1919 error
= vdev_fault(spa
, zc
->zc_guid
, zc
->zc_obj
);
1922 case VDEV_STATE_DEGRADED
:
1923 if (zc
->zc_obj
!= VDEV_AUX_ERR_EXCEEDED
&&
1924 zc
->zc_obj
!= VDEV_AUX_EXTERNAL
)
1925 zc
->zc_obj
= VDEV_AUX_ERR_EXCEEDED
;
1927 error
= vdev_degrade(spa
, zc
->zc_guid
, zc
->zc_obj
);
1931 error
= SET_ERROR(EINVAL
);
1933 zc
->zc_cookie
= newstate
;
1934 spa_close(spa
, FTAG
);
1939 zfs_ioc_vdev_attach(zfs_cmd_t
*zc
)
1942 int replacing
= zc
->zc_cookie
;
1946 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1949 if ((error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1950 zc
->zc_iflags
, &config
)) == 0) {
1951 error
= spa_vdev_attach(spa
, zc
->zc_guid
, config
, replacing
);
1952 nvlist_free(config
);
1955 spa_close(spa
, FTAG
);
1960 zfs_ioc_vdev_detach(zfs_cmd_t
*zc
)
1965 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1968 error
= spa_vdev_detach(spa
, zc
->zc_guid
, 0, B_FALSE
);
1970 spa_close(spa
, FTAG
);
1975 zfs_ioc_vdev_split(zfs_cmd_t
*zc
)
1978 nvlist_t
*config
, *props
= NULL
;
1980 boolean_t exp
= !!(zc
->zc_cookie
& ZPOOL_EXPORT_AFTER_SPLIT
);
1982 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
1985 if ((error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
1986 zc
->zc_iflags
, &config
))) {
1987 spa_close(spa
, FTAG
);
1991 if (zc
->zc_nvlist_src_size
!= 0 && (error
=
1992 get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
1993 zc
->zc_iflags
, &props
))) {
1994 spa_close(spa
, FTAG
);
1995 nvlist_free(config
);
1999 error
= spa_vdev_split_mirror(spa
, zc
->zc_string
, config
, props
, exp
);
2001 spa_close(spa
, FTAG
);
2003 nvlist_free(config
);
2010 zfs_ioc_vdev_setpath(zfs_cmd_t
*zc
)
2013 char *path
= zc
->zc_value
;
2014 uint64_t guid
= zc
->zc_guid
;
2017 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
2021 error
= spa_vdev_setpath(spa
, guid
, path
);
2022 spa_close(spa
, FTAG
);
2027 zfs_ioc_vdev_setfru(zfs_cmd_t
*zc
)
2030 char *fru
= zc
->zc_value
;
2031 uint64_t guid
= zc
->zc_guid
;
2034 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
2038 error
= spa_vdev_setfru(spa
, guid
, fru
);
2039 spa_close(spa
, FTAG
);
2044 zfs_ioc_objset_stats_impl(zfs_cmd_t
*zc
, objset_t
*os
)
2049 dmu_objset_fast_stat(os
, &zc
->zc_objset_stats
);
2051 if (zc
->zc_nvlist_dst
!= 0 &&
2052 (error
= dsl_prop_get_all(os
, &nv
)) == 0) {
2053 dmu_objset_stats(os
, nv
);
2055 * NB: zvol_get_stats() will read the objset contents,
2056 * which we aren't supposed to do with a
2057 * DS_MODE_USER hold, because it could be
2058 * inconsistent. So this is a bit of a workaround...
2059 * XXX reading without owning
2061 if (!zc
->zc_objset_stats
.dds_inconsistent
&&
2062 dmu_objset_type(os
) == DMU_OST_ZVOL
) {
2063 error
= zvol_get_stats(os
, nv
);
2071 error
= put_nvlist(zc
, nv
);
2080 * zc_name name of filesystem
2081 * zc_nvlist_dst_size size of buffer for property nvlist
2084 * zc_objset_stats stats
2085 * zc_nvlist_dst property nvlist
2086 * zc_nvlist_dst_size size of property nvlist
2089 zfs_ioc_objset_stats(zfs_cmd_t
*zc
)
2094 error
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
);
2096 error
= zfs_ioc_objset_stats_impl(zc
, os
);
2097 dmu_objset_rele(os
, FTAG
);
2105 * zc_name name of filesystem
2106 * zc_nvlist_dst_size size of buffer for property nvlist
2109 * zc_nvlist_dst received property nvlist
2110 * zc_nvlist_dst_size size of received property nvlist
2112 * Gets received properties (distinct from local properties on or after
2113 * SPA_VERSION_RECVD_PROPS) for callers who want to differentiate received from
2114 * local property values.
2117 zfs_ioc_objset_recvd_props(zfs_cmd_t
*zc
)
2123 * Without this check, we would return local property values if the
2124 * caller has not already received properties on or after
2125 * SPA_VERSION_RECVD_PROPS.
2127 if (!dsl_prop_get_hasrecvd(zc
->zc_name
))
2128 return (SET_ERROR(ENOTSUP
));
2130 if (zc
->zc_nvlist_dst
!= 0 &&
2131 (error
= dsl_prop_get_received(zc
->zc_name
, &nv
)) == 0) {
2132 error
= put_nvlist(zc
, nv
);
2140 nvl_add_zplprop(objset_t
*os
, nvlist_t
*props
, zfs_prop_t prop
)
2146 * zfs_get_zplprop() will either find a value or give us
2147 * the default value (if there is one).
2149 if ((error
= zfs_get_zplprop(os
, prop
, &value
)) != 0)
2151 VERIFY(nvlist_add_uint64(props
, zfs_prop_to_name(prop
), value
) == 0);
2157 * zc_name name of filesystem
2158 * zc_nvlist_dst_size size of buffer for zpl property nvlist
2161 * zc_nvlist_dst zpl property nvlist
2162 * zc_nvlist_dst_size size of zpl property nvlist
2165 zfs_ioc_objset_zplprops(zfs_cmd_t
*zc
)
2170 /* XXX reading without owning */
2171 if ((err
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
)))
2174 dmu_objset_fast_stat(os
, &zc
->zc_objset_stats
);
2177 * NB: nvl_add_zplprop() will read the objset contents,
2178 * which we aren't supposed to do with a DS_MODE_USER
2179 * hold, because it could be inconsistent.
2181 if (zc
->zc_nvlist_dst
!= 0 &&
2182 !zc
->zc_objset_stats
.dds_inconsistent
&&
2183 dmu_objset_type(os
) == DMU_OST_ZFS
) {
2186 VERIFY(nvlist_alloc(&nv
, NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
2187 if ((err
= nvl_add_zplprop(os
, nv
, ZFS_PROP_VERSION
)) == 0 &&
2188 (err
= nvl_add_zplprop(os
, nv
, ZFS_PROP_NORMALIZE
)) == 0 &&
2189 (err
= nvl_add_zplprop(os
, nv
, ZFS_PROP_UTF8ONLY
)) == 0 &&
2190 (err
= nvl_add_zplprop(os
, nv
, ZFS_PROP_CASE
)) == 0)
2191 err
= put_nvlist(zc
, nv
);
2194 err
= SET_ERROR(ENOENT
);
2196 dmu_objset_rele(os
, FTAG
);
2202 * zc_name name of filesystem
2203 * zc_cookie zap cursor
2204 * zc_nvlist_dst_size size of buffer for property nvlist
2207 * zc_name name of next filesystem
2208 * zc_cookie zap cursor
2209 * zc_objset_stats stats
2210 * zc_nvlist_dst property nvlist
2211 * zc_nvlist_dst_size size of property nvlist
2214 zfs_ioc_dataset_list_next(zfs_cmd_t
*zc
)
2219 size_t orig_len
= strlen(zc
->zc_name
);
2222 if ((error
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
))) {
2223 if (error
== ENOENT
)
2224 error
= SET_ERROR(ESRCH
);
2228 p
= strrchr(zc
->zc_name
, '/');
2229 if (p
== NULL
|| p
[1] != '\0')
2230 (void) strlcat(zc
->zc_name
, "/", sizeof (zc
->zc_name
));
2231 p
= zc
->zc_name
+ strlen(zc
->zc_name
);
2234 error
= dmu_dir_list_next(os
,
2235 sizeof (zc
->zc_name
) - (p
- zc
->zc_name
), p
,
2236 NULL
, &zc
->zc_cookie
);
2237 if (error
== ENOENT
)
2238 error
= SET_ERROR(ESRCH
);
2239 } while (error
== 0 && zfs_dataset_name_hidden(zc
->zc_name
));
2240 dmu_objset_rele(os
, FTAG
);
2243 * If it's an internal dataset (ie. with a '$' in its name),
2244 * don't try to get stats for it, otherwise we'll return ENOENT.
2246 if (error
== 0 && strchr(zc
->zc_name
, '$') == NULL
) {
2247 error
= zfs_ioc_objset_stats(zc
); /* fill in the stats */
2248 if (error
== ENOENT
) {
2249 /* We lost a race with destroy, get the next one. */
2250 zc
->zc_name
[orig_len
] = '\0';
2259 * zc_name name of filesystem
2260 * zc_cookie zap cursor
2261 * zc_nvlist_src iteration range nvlist
2262 * zc_nvlist_src_size size of iteration range nvlist
2265 * zc_name name of next snapshot
2266 * zc_objset_stats stats
2267 * zc_nvlist_dst property nvlist
2268 * zc_nvlist_dst_size size of property nvlist
2271 zfs_ioc_snapshot_list_next(zfs_cmd_t
*zc
)
2274 objset_t
*os
, *ossnap
;
2276 uint64_t min_txg
= 0, max_txg
= 0;
2278 if (zc
->zc_nvlist_src_size
!= 0) {
2279 nvlist_t
*props
= NULL
;
2280 error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
2281 zc
->zc_iflags
, &props
);
2284 (void) nvlist_lookup_uint64(props
, SNAP_ITER_MIN_TXG
,
2286 (void) nvlist_lookup_uint64(props
, SNAP_ITER_MAX_TXG
,
2291 error
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
);
2293 return (error
== ENOENT
? ESRCH
: error
);
2297 * A dataset name of maximum length cannot have any snapshots,
2298 * so exit immediately.
2300 if (strlcat(zc
->zc_name
, "@", sizeof (zc
->zc_name
)) >=
2301 ZFS_MAX_DATASET_NAME_LEN
) {
2302 dmu_objset_rele(os
, FTAG
);
2303 return (SET_ERROR(ESRCH
));
2306 while (error
== 0) {
2307 if (issig(JUSTLOOKING
) && issig(FORREAL
)) {
2308 error
= SET_ERROR(EINTR
);
2312 error
= dmu_snapshot_list_next(os
,
2313 sizeof (zc
->zc_name
) - strlen(zc
->zc_name
),
2314 zc
->zc_name
+ strlen(zc
->zc_name
), &zc
->zc_obj
,
2315 &zc
->zc_cookie
, NULL
);
2316 if (error
== ENOENT
) {
2317 error
= SET_ERROR(ESRCH
);
2319 } else if (error
!= 0) {
2323 error
= dsl_dataset_hold_obj(dmu_objset_pool(os
), zc
->zc_obj
,
2328 if ((min_txg
!= 0 && dsl_get_creationtxg(ds
) < min_txg
) ||
2329 (max_txg
!= 0 && dsl_get_creationtxg(ds
) > max_txg
)) {
2330 dsl_dataset_rele(ds
, FTAG
);
2331 /* undo snapshot name append */
2332 *(strchr(zc
->zc_name
, '@') + 1) = '\0';
2337 if (zc
->zc_simple
) {
2338 dsl_dataset_rele(ds
, FTAG
);
2342 if ((error
= dmu_objset_from_ds(ds
, &ossnap
)) != 0) {
2343 dsl_dataset_rele(ds
, FTAG
);
2346 if ((error
= zfs_ioc_objset_stats_impl(zc
, ossnap
)) != 0) {
2347 dsl_dataset_rele(ds
, FTAG
);
2350 dsl_dataset_rele(ds
, FTAG
);
2354 dmu_objset_rele(os
, FTAG
);
2355 /* if we failed, undo the @ that we tacked on to zc_name */
2357 *strchr(zc
->zc_name
, '@') = '\0';
2362 zfs_prop_set_userquota(const char *dsname
, nvpair_t
*pair
)
2364 const char *propname
= nvpair_name(pair
);
2366 unsigned int vallen
;
2369 zfs_userquota_prop_t type
;
2375 if (nvpair_type(pair
) == DATA_TYPE_NVLIST
) {
2377 VERIFY(nvpair_value_nvlist(pair
, &attrs
) == 0);
2378 if (nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
2380 return (SET_ERROR(EINVAL
));
2384 * A correctly constructed propname is encoded as
2385 * userquota@<rid>-<domain>.
2387 if ((dash
= strchr(propname
, '-')) == NULL
||
2388 nvpair_value_uint64_array(pair
, &valary
, &vallen
) != 0 ||
2390 return (SET_ERROR(EINVAL
));
2397 err
= zfsvfs_hold(dsname
, FTAG
, &zfsvfs
, B_FALSE
);
2399 err
= zfs_set_userquota(zfsvfs
, type
, domain
, rid
, quota
);
2400 zfsvfs_rele(zfsvfs
, FTAG
);
2407 * If the named property is one that has a special function to set its value,
2408 * return 0 on success and a positive error code on failure; otherwise if it is
2409 * not one of the special properties handled by this function, return -1.
2411 * XXX: It would be better for callers of the property interface if we handled
2412 * these special cases in dsl_prop.c (in the dsl layer).
2415 zfs_prop_set_special(const char *dsname
, zprop_source_t source
,
2418 const char *propname
= nvpair_name(pair
);
2419 zfs_prop_t prop
= zfs_name_to_prop(propname
);
2420 uint64_t intval
= 0;
2421 char *strval
= NULL
;
2424 if (prop
== ZPROP_INVAL
) {
2425 if (zfs_prop_userquota(propname
))
2426 return (zfs_prop_set_userquota(dsname
, pair
));
2430 if (nvpair_type(pair
) == DATA_TYPE_NVLIST
) {
2432 VERIFY(nvpair_value_nvlist(pair
, &attrs
) == 0);
2433 VERIFY(nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
2437 /* all special properties are numeric except for keylocation */
2438 if (zfs_prop_get_type(prop
) == PROP_TYPE_STRING
) {
2439 strval
= fnvpair_value_string(pair
);
2441 intval
= fnvpair_value_uint64(pair
);
2445 case ZFS_PROP_QUOTA
:
2446 err
= dsl_dir_set_quota(dsname
, source
, intval
);
2448 case ZFS_PROP_REFQUOTA
:
2449 err
= dsl_dataset_set_refquota(dsname
, source
, intval
);
2451 case ZFS_PROP_FILESYSTEM_LIMIT
:
2452 case ZFS_PROP_SNAPSHOT_LIMIT
:
2453 if (intval
== UINT64_MAX
) {
2454 /* clearing the limit, just do it */
2457 err
= dsl_dir_activate_fs_ss_limit(dsname
);
2460 * Set err to -1 to force the zfs_set_prop_nvlist code down the
2461 * default path to set the value in the nvlist.
2466 case ZFS_PROP_KEYLOCATION
:
2467 err
= dsl_crypto_can_set_keylocation(dsname
, strval
);
2470 * Set err to -1 to force the zfs_set_prop_nvlist code down the
2471 * default path to set the value in the nvlist.
2476 case ZFS_PROP_RESERVATION
:
2477 err
= dsl_dir_set_reservation(dsname
, source
, intval
);
2479 case ZFS_PROP_REFRESERVATION
:
2480 err
= dsl_dataset_set_refreservation(dsname
, source
, intval
);
2482 case ZFS_PROP_VOLSIZE
:
2483 err
= zvol_set_volsize(dsname
, intval
);
2485 case ZFS_PROP_SNAPDEV
:
2486 err
= zvol_set_snapdev(dsname
, source
, intval
);
2488 case ZFS_PROP_VOLMODE
:
2489 err
= zvol_set_volmode(dsname
, source
, intval
);
2491 case ZFS_PROP_VERSION
:
2495 if ((err
= zfsvfs_hold(dsname
, FTAG
, &zfsvfs
, B_TRUE
)) != 0)
2498 err
= zfs_set_version(zfsvfs
, intval
);
2499 zfsvfs_rele(zfsvfs
, FTAG
);
2501 if (err
== 0 && intval
>= ZPL_VERSION_USERSPACE
) {
2504 zc
= kmem_zalloc(sizeof (zfs_cmd_t
), KM_SLEEP
);
2505 (void) strcpy(zc
->zc_name
, dsname
);
2506 (void) zfs_ioc_userspace_upgrade(zc
);
2507 (void) zfs_ioc_id_quota_upgrade(zc
);
2508 kmem_free(zc
, sizeof (zfs_cmd_t
));
2520 * This function is best effort. If it fails to set any of the given properties,
2521 * it continues to set as many as it can and returns the last error
2522 * encountered. If the caller provides a non-NULL errlist, it will be filled in
2523 * with the list of names of all the properties that failed along with the
2524 * corresponding error numbers.
2526 * If every property is set successfully, zero is returned and errlist is not
2530 zfs_set_prop_nvlist(const char *dsname
, zprop_source_t source
, nvlist_t
*nvl
,
2539 nvlist_t
*genericnvl
= fnvlist_alloc();
2540 nvlist_t
*retrynvl
= fnvlist_alloc();
2543 while ((pair
= nvlist_next_nvpair(nvl
, pair
)) != NULL
) {
2544 const char *propname
= nvpair_name(pair
);
2545 zfs_prop_t prop
= zfs_name_to_prop(propname
);
2548 /* decode the property value */
2550 if (nvpair_type(pair
) == DATA_TYPE_NVLIST
) {
2552 attrs
= fnvpair_value_nvlist(pair
);
2553 if (nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
2555 err
= SET_ERROR(EINVAL
);
2558 /* Validate value type */
2559 if (err
== 0 && source
== ZPROP_SRC_INHERITED
) {
2560 /* inherited properties are expected to be booleans */
2561 if (nvpair_type(propval
) != DATA_TYPE_BOOLEAN
)
2562 err
= SET_ERROR(EINVAL
);
2563 } else if (err
== 0 && prop
== ZPROP_INVAL
) {
2564 if (zfs_prop_user(propname
)) {
2565 if (nvpair_type(propval
) != DATA_TYPE_STRING
)
2566 err
= SET_ERROR(EINVAL
);
2567 } else if (zfs_prop_userquota(propname
)) {
2568 if (nvpair_type(propval
) !=
2569 DATA_TYPE_UINT64_ARRAY
)
2570 err
= SET_ERROR(EINVAL
);
2572 err
= SET_ERROR(EINVAL
);
2574 } else if (err
== 0) {
2575 if (nvpair_type(propval
) == DATA_TYPE_STRING
) {
2576 if (zfs_prop_get_type(prop
) != PROP_TYPE_STRING
)
2577 err
= SET_ERROR(EINVAL
);
2578 } else if (nvpair_type(propval
) == DATA_TYPE_UINT64
) {
2581 intval
= fnvpair_value_uint64(propval
);
2583 switch (zfs_prop_get_type(prop
)) {
2584 case PROP_TYPE_NUMBER
:
2586 case PROP_TYPE_STRING
:
2587 err
= SET_ERROR(EINVAL
);
2589 case PROP_TYPE_INDEX
:
2590 if (zfs_prop_index_to_string(prop
,
2591 intval
, &unused
) != 0)
2592 err
= SET_ERROR(EINVAL
);
2596 "unknown property type");
2599 err
= SET_ERROR(EINVAL
);
2603 /* Validate permissions */
2605 err
= zfs_check_settable(dsname
, pair
, CRED());
2608 if (source
== ZPROP_SRC_INHERITED
)
2609 err
= -1; /* does not need special handling */
2611 err
= zfs_prop_set_special(dsname
, source
,
2615 * For better performance we build up a list of
2616 * properties to set in a single transaction.
2618 err
= nvlist_add_nvpair(genericnvl
, pair
);
2619 } else if (err
!= 0 && nvl
!= retrynvl
) {
2621 * This may be a spurious error caused by
2622 * receiving quota and reservation out of order.
2623 * Try again in a second pass.
2625 err
= nvlist_add_nvpair(retrynvl
, pair
);
2630 if (errlist
!= NULL
)
2631 fnvlist_add_int32(errlist
, propname
, err
);
2636 if (nvl
!= retrynvl
&& !nvlist_empty(retrynvl
)) {
2641 if (!nvlist_empty(genericnvl
) &&
2642 dsl_props_set(dsname
, source
, genericnvl
) != 0) {
2644 * If this fails, we still want to set as many properties as we
2645 * can, so try setting them individually.
2648 while ((pair
= nvlist_next_nvpair(genericnvl
, pair
)) != NULL
) {
2649 const char *propname
= nvpair_name(pair
);
2653 if (nvpair_type(pair
) == DATA_TYPE_NVLIST
) {
2655 attrs
= fnvpair_value_nvlist(pair
);
2656 propval
= fnvlist_lookup_nvpair(attrs
,
2660 if (nvpair_type(propval
) == DATA_TYPE_STRING
) {
2661 strval
= fnvpair_value_string(propval
);
2662 err
= dsl_prop_set_string(dsname
, propname
,
2664 } else if (nvpair_type(propval
) == DATA_TYPE_BOOLEAN
) {
2665 err
= dsl_prop_inherit(dsname
, propname
,
2668 intval
= fnvpair_value_uint64(propval
);
2669 err
= dsl_prop_set_int(dsname
, propname
, source
,
2674 if (errlist
!= NULL
) {
2675 fnvlist_add_int32(errlist
, propname
,
2682 nvlist_free(genericnvl
);
2683 nvlist_free(retrynvl
);
2689 * Check that all the properties are valid user properties.
2692 zfs_check_userprops(nvlist_t
*nvl
)
2694 nvpair_t
*pair
= NULL
;
2696 while ((pair
= nvlist_next_nvpair(nvl
, pair
)) != NULL
) {
2697 const char *propname
= nvpair_name(pair
);
2699 if (!zfs_prop_user(propname
) ||
2700 nvpair_type(pair
) != DATA_TYPE_STRING
)
2701 return (SET_ERROR(EINVAL
));
2703 if (strlen(propname
) >= ZAP_MAXNAMELEN
)
2704 return (SET_ERROR(ENAMETOOLONG
));
2706 if (strlen(fnvpair_value_string(pair
)) >= ZAP_MAXVALUELEN
)
2707 return (SET_ERROR(E2BIG
));
2713 props_skip(nvlist_t
*props
, nvlist_t
*skipped
, nvlist_t
**newprops
)
2717 VERIFY(nvlist_alloc(newprops
, NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
2720 while ((pair
= nvlist_next_nvpair(props
, pair
)) != NULL
) {
2721 if (nvlist_exists(skipped
, nvpair_name(pair
)))
2724 VERIFY(nvlist_add_nvpair(*newprops
, pair
) == 0);
2729 clear_received_props(const char *dsname
, nvlist_t
*props
,
2733 nvlist_t
*cleared_props
= NULL
;
2734 props_skip(props
, skipped
, &cleared_props
);
2735 if (!nvlist_empty(cleared_props
)) {
2737 * Acts on local properties until the dataset has received
2738 * properties at least once on or after SPA_VERSION_RECVD_PROPS.
2740 zprop_source_t flags
= (ZPROP_SRC_NONE
|
2741 (dsl_prop_get_hasrecvd(dsname
) ? ZPROP_SRC_RECEIVED
: 0));
2742 err
= zfs_set_prop_nvlist(dsname
, flags
, cleared_props
, NULL
);
2744 nvlist_free(cleared_props
);
2750 * zc_name name of filesystem
2751 * zc_value name of property to set
2752 * zc_nvlist_src{_size} nvlist of properties to apply
2753 * zc_cookie received properties flag
2756 * zc_nvlist_dst{_size} error for each unapplied received property
2759 zfs_ioc_set_prop(zfs_cmd_t
*zc
)
2762 boolean_t received
= zc
->zc_cookie
;
2763 zprop_source_t source
= (received
? ZPROP_SRC_RECEIVED
:
2768 if ((error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
2769 zc
->zc_iflags
, &nvl
)) != 0)
2773 nvlist_t
*origprops
;
2775 if (dsl_prop_get_received(zc
->zc_name
, &origprops
) == 0) {
2776 (void) clear_received_props(zc
->zc_name
,
2778 nvlist_free(origprops
);
2781 error
= dsl_prop_set_hasrecvd(zc
->zc_name
);
2784 errors
= fnvlist_alloc();
2786 error
= zfs_set_prop_nvlist(zc
->zc_name
, source
, nvl
, errors
);
2788 if (zc
->zc_nvlist_dst
!= 0 && errors
!= NULL
) {
2789 (void) put_nvlist(zc
, errors
);
2792 nvlist_free(errors
);
2799 * zc_name name of filesystem
2800 * zc_value name of property to inherit
2801 * zc_cookie revert to received value if TRUE
2806 zfs_ioc_inherit_prop(zfs_cmd_t
*zc
)
2808 const char *propname
= zc
->zc_value
;
2809 zfs_prop_t prop
= zfs_name_to_prop(propname
);
2810 boolean_t received
= zc
->zc_cookie
;
2811 zprop_source_t source
= (received
2812 ? ZPROP_SRC_NONE
/* revert to received value, if any */
2813 : ZPROP_SRC_INHERITED
); /* explicitly inherit */
2821 * Only check this in the non-received case. We want to allow
2822 * 'inherit -S' to revert non-inheritable properties like quota
2823 * and reservation to the received or default values even though
2824 * they are not considered inheritable.
2826 if (prop
!= ZPROP_INVAL
&& !zfs_prop_inheritable(prop
))
2827 return (SET_ERROR(EINVAL
));
2830 if (prop
== ZPROP_INVAL
) {
2831 if (!zfs_prop_user(propname
))
2832 return (SET_ERROR(EINVAL
));
2834 type
= PROP_TYPE_STRING
;
2835 } else if (prop
== ZFS_PROP_VOLSIZE
|| prop
== ZFS_PROP_VERSION
) {
2836 return (SET_ERROR(EINVAL
));
2838 type
= zfs_prop_get_type(prop
);
2842 * zfs_prop_set_special() expects properties in the form of an
2843 * nvpair with type info.
2845 dummy
= fnvlist_alloc();
2848 case PROP_TYPE_STRING
:
2849 VERIFY(0 == nvlist_add_string(dummy
, propname
, ""));
2851 case PROP_TYPE_NUMBER
:
2852 case PROP_TYPE_INDEX
:
2853 VERIFY(0 == nvlist_add_uint64(dummy
, propname
, 0));
2856 err
= SET_ERROR(EINVAL
);
2860 pair
= nvlist_next_nvpair(dummy
, NULL
);
2862 err
= SET_ERROR(EINVAL
);
2864 err
= zfs_prop_set_special(zc
->zc_name
, source
, pair
);
2865 if (err
== -1) /* property is not "special", needs handling */
2866 err
= dsl_prop_inherit(zc
->zc_name
, zc
->zc_value
,
2876 zfs_ioc_pool_set_props(zfs_cmd_t
*zc
)
2883 if ((error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
2884 zc
->zc_iflags
, &props
)))
2888 * If the only property is the configfile, then just do a spa_lookup()
2889 * to handle the faulted case.
2891 pair
= nvlist_next_nvpair(props
, NULL
);
2892 if (pair
!= NULL
&& strcmp(nvpair_name(pair
),
2893 zpool_prop_to_name(ZPOOL_PROP_CACHEFILE
)) == 0 &&
2894 nvlist_next_nvpair(props
, pair
) == NULL
) {
2895 mutex_enter(&spa_namespace_lock
);
2896 if ((spa
= spa_lookup(zc
->zc_name
)) != NULL
) {
2897 spa_configfile_set(spa
, props
, B_FALSE
);
2898 spa_write_cachefile(spa
, B_FALSE
, B_TRUE
);
2900 mutex_exit(&spa_namespace_lock
);
2907 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0) {
2912 error
= spa_prop_set(spa
, props
);
2915 spa_close(spa
, FTAG
);
2921 zfs_ioc_pool_get_props(zfs_cmd_t
*zc
)
2925 nvlist_t
*nvp
= NULL
;
2927 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0) {
2929 * If the pool is faulted, there may be properties we can still
2930 * get (such as altroot and cachefile), so attempt to get them
2933 mutex_enter(&spa_namespace_lock
);
2934 if ((spa
= spa_lookup(zc
->zc_name
)) != NULL
)
2935 error
= spa_prop_get(spa
, &nvp
);
2936 mutex_exit(&spa_namespace_lock
);
2938 error
= spa_prop_get(spa
, &nvp
);
2939 spa_close(spa
, FTAG
);
2942 if (error
== 0 && zc
->zc_nvlist_dst
!= 0)
2943 error
= put_nvlist(zc
, nvp
);
2945 error
= SET_ERROR(EFAULT
);
2953 * zc_name name of filesystem
2954 * zc_nvlist_src{_size} nvlist of delegated permissions
2955 * zc_perm_action allow/unallow flag
2960 zfs_ioc_set_fsacl(zfs_cmd_t
*zc
)
2963 nvlist_t
*fsaclnv
= NULL
;
2965 if ((error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
2966 zc
->zc_iflags
, &fsaclnv
)) != 0)
2970 * Verify nvlist is constructed correctly
2972 if ((error
= zfs_deleg_verify_nvlist(fsaclnv
)) != 0) {
2973 nvlist_free(fsaclnv
);
2974 return (SET_ERROR(EINVAL
));
2978 * If we don't have PRIV_SYS_MOUNT, then validate
2979 * that user is allowed to hand out each permission in
2983 error
= secpolicy_zfs(CRED());
2985 if (zc
->zc_perm_action
== B_FALSE
) {
2986 error
= dsl_deleg_can_allow(zc
->zc_name
,
2989 error
= dsl_deleg_can_unallow(zc
->zc_name
,
2995 error
= dsl_deleg_set(zc
->zc_name
, fsaclnv
, zc
->zc_perm_action
);
2997 nvlist_free(fsaclnv
);
3003 * zc_name name of filesystem
3006 * zc_nvlist_src{_size} nvlist of delegated permissions
3009 zfs_ioc_get_fsacl(zfs_cmd_t
*zc
)
3014 if ((error
= dsl_deleg_get(zc
->zc_name
, &nvp
)) == 0) {
3015 error
= put_nvlist(zc
, nvp
);
3024 zfs_create_cb(objset_t
*os
, void *arg
, cred_t
*cr
, dmu_tx_t
*tx
)
3026 zfs_creat_t
*zct
= arg
;
3028 zfs_create_fs(os
, cr
, zct
->zct_zplprops
, tx
);
3031 #define ZFS_PROP_UNDEFINED ((uint64_t)-1)
3035 * os parent objset pointer (NULL if root fs)
3036 * fuids_ok fuids allowed in this version of the spa?
3037 * sa_ok SAs allowed in this version of the spa?
3038 * createprops list of properties requested by creator
3041 * zplprops values for the zplprops we attach to the master node object
3042 * is_ci true if requested file system will be purely case-insensitive
3044 * Determine the settings for utf8only, normalization and
3045 * casesensitivity. Specific values may have been requested by the
3046 * creator and/or we can inherit values from the parent dataset. If
3047 * the file system is of too early a vintage, a creator can not
3048 * request settings for these properties, even if the requested
3049 * setting is the default value. We don't actually want to create dsl
3050 * properties for these, so remove them from the source nvlist after
3054 zfs_fill_zplprops_impl(objset_t
*os
, uint64_t zplver
,
3055 boolean_t fuids_ok
, boolean_t sa_ok
, nvlist_t
*createprops
,
3056 nvlist_t
*zplprops
, boolean_t
*is_ci
)
3058 uint64_t sense
= ZFS_PROP_UNDEFINED
;
3059 uint64_t norm
= ZFS_PROP_UNDEFINED
;
3060 uint64_t u8
= ZFS_PROP_UNDEFINED
;
3063 ASSERT(zplprops
!= NULL
);
3065 /* parent dataset must be a filesystem */
3066 if (os
!= NULL
&& os
->os_phys
->os_type
!= DMU_OST_ZFS
)
3067 return (SET_ERROR(ZFS_ERR_WRONG_PARENT
));
3070 * Pull out creator prop choices, if any.
3073 (void) nvlist_lookup_uint64(createprops
,
3074 zfs_prop_to_name(ZFS_PROP_VERSION
), &zplver
);
3075 (void) nvlist_lookup_uint64(createprops
,
3076 zfs_prop_to_name(ZFS_PROP_NORMALIZE
), &norm
);
3077 (void) nvlist_remove_all(createprops
,
3078 zfs_prop_to_name(ZFS_PROP_NORMALIZE
));
3079 (void) nvlist_lookup_uint64(createprops
,
3080 zfs_prop_to_name(ZFS_PROP_UTF8ONLY
), &u8
);
3081 (void) nvlist_remove_all(createprops
,
3082 zfs_prop_to_name(ZFS_PROP_UTF8ONLY
));
3083 (void) nvlist_lookup_uint64(createprops
,
3084 zfs_prop_to_name(ZFS_PROP_CASE
), &sense
);
3085 (void) nvlist_remove_all(createprops
,
3086 zfs_prop_to_name(ZFS_PROP_CASE
));
3090 * If the zpl version requested is whacky or the file system
3091 * or pool is version is too "young" to support normalization
3092 * and the creator tried to set a value for one of the props,
3095 if ((zplver
< ZPL_VERSION_INITIAL
|| zplver
> ZPL_VERSION
) ||
3096 (zplver
>= ZPL_VERSION_FUID
&& !fuids_ok
) ||
3097 (zplver
>= ZPL_VERSION_SA
&& !sa_ok
) ||
3098 (zplver
< ZPL_VERSION_NORMALIZATION
&&
3099 (norm
!= ZFS_PROP_UNDEFINED
|| u8
!= ZFS_PROP_UNDEFINED
||
3100 sense
!= ZFS_PROP_UNDEFINED
)))
3101 return (SET_ERROR(ENOTSUP
));
3104 * Put the version in the zplprops
3106 VERIFY(nvlist_add_uint64(zplprops
,
3107 zfs_prop_to_name(ZFS_PROP_VERSION
), zplver
) == 0);
3109 if (norm
== ZFS_PROP_UNDEFINED
&&
3110 (error
= zfs_get_zplprop(os
, ZFS_PROP_NORMALIZE
, &norm
)) != 0)
3112 VERIFY(nvlist_add_uint64(zplprops
,
3113 zfs_prop_to_name(ZFS_PROP_NORMALIZE
), norm
) == 0);
3116 * If we're normalizing, names must always be valid UTF-8 strings.
3120 if (u8
== ZFS_PROP_UNDEFINED
&&
3121 (error
= zfs_get_zplprop(os
, ZFS_PROP_UTF8ONLY
, &u8
)) != 0)
3123 VERIFY(nvlist_add_uint64(zplprops
,
3124 zfs_prop_to_name(ZFS_PROP_UTF8ONLY
), u8
) == 0);
3126 if (sense
== ZFS_PROP_UNDEFINED
&&
3127 (error
= zfs_get_zplprop(os
, ZFS_PROP_CASE
, &sense
)) != 0)
3129 VERIFY(nvlist_add_uint64(zplprops
,
3130 zfs_prop_to_name(ZFS_PROP_CASE
), sense
) == 0);
3133 *is_ci
= (sense
== ZFS_CASE_INSENSITIVE
);
3139 zfs_fill_zplprops(const char *dataset
, nvlist_t
*createprops
,
3140 nvlist_t
*zplprops
, boolean_t
*is_ci
)
3142 boolean_t fuids_ok
, sa_ok
;
3143 uint64_t zplver
= ZPL_VERSION
;
3144 objset_t
*os
= NULL
;
3145 char parentname
[ZFS_MAX_DATASET_NAME_LEN
];
3150 zfs_get_parent(dataset
, parentname
, sizeof (parentname
));
3152 if ((error
= spa_open(dataset
, &spa
, FTAG
)) != 0)
3155 spa_vers
= spa_version(spa
);
3156 spa_close(spa
, FTAG
);
3158 zplver
= zfs_zpl_version_map(spa_vers
);
3159 fuids_ok
= (zplver
>= ZPL_VERSION_FUID
);
3160 sa_ok
= (zplver
>= ZPL_VERSION_SA
);
3163 * Open parent object set so we can inherit zplprop values.
3165 if ((error
= dmu_objset_hold(parentname
, FTAG
, &os
)) != 0)
3168 error
= zfs_fill_zplprops_impl(os
, zplver
, fuids_ok
, sa_ok
, createprops
,
3170 dmu_objset_rele(os
, FTAG
);
3175 zfs_fill_zplprops_root(uint64_t spa_vers
, nvlist_t
*createprops
,
3176 nvlist_t
*zplprops
, boolean_t
*is_ci
)
3180 uint64_t zplver
= ZPL_VERSION
;
3183 zplver
= zfs_zpl_version_map(spa_vers
);
3184 fuids_ok
= (zplver
>= ZPL_VERSION_FUID
);
3185 sa_ok
= (zplver
>= ZPL_VERSION_SA
);
3187 error
= zfs_fill_zplprops_impl(NULL
, zplver
, fuids_ok
, sa_ok
,
3188 createprops
, zplprops
, is_ci
);
3194 * "type" -> dmu_objset_type_t (int32)
3195 * (optional) "props" -> { prop -> value }
3196 * (optional) "hidden_args" -> { "wkeydata" -> value }
3197 * raw uint8_t array of encryption wrapping key data (32 bytes)
3200 * outnvl: propname -> error code (int32)
3203 static const zfs_ioc_key_t zfs_keys_create
[] = {
3204 {"type", DATA_TYPE_INT32
, 0},
3205 {"props", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
3206 {"hidden_args", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
3210 zfs_ioc_create(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3213 zfs_creat_t zct
= { 0 };
3214 nvlist_t
*nvprops
= NULL
;
3215 nvlist_t
*hidden_args
= NULL
;
3216 void (*cbfunc
)(objset_t
*os
, void *arg
, cred_t
*cr
, dmu_tx_t
*tx
);
3217 dmu_objset_type_t type
;
3218 boolean_t is_insensitive
= B_FALSE
;
3219 dsl_crypto_params_t
*dcp
= NULL
;
3221 type
= (dmu_objset_type_t
)fnvlist_lookup_int32(innvl
, "type");
3222 (void) nvlist_lookup_nvlist(innvl
, "props", &nvprops
);
3223 (void) nvlist_lookup_nvlist(innvl
, ZPOOL_HIDDEN_ARGS
, &hidden_args
);
3227 cbfunc
= zfs_create_cb
;
3231 cbfunc
= zvol_create_cb
;
3238 if (strchr(fsname
, '@') ||
3239 strchr(fsname
, '%'))
3240 return (SET_ERROR(EINVAL
));
3242 zct
.zct_props
= nvprops
;
3245 return (SET_ERROR(EINVAL
));
3247 if (type
== DMU_OST_ZVOL
) {
3248 uint64_t volsize
, volblocksize
;
3250 if (nvprops
== NULL
)
3251 return (SET_ERROR(EINVAL
));
3252 if (nvlist_lookup_uint64(nvprops
,
3253 zfs_prop_to_name(ZFS_PROP_VOLSIZE
), &volsize
) != 0)
3254 return (SET_ERROR(EINVAL
));
3256 if ((error
= nvlist_lookup_uint64(nvprops
,
3257 zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE
),
3258 &volblocksize
)) != 0 && error
!= ENOENT
)
3259 return (SET_ERROR(EINVAL
));
3262 volblocksize
= zfs_prop_default_numeric(
3263 ZFS_PROP_VOLBLOCKSIZE
);
3265 if ((error
= zvol_check_volblocksize(fsname
,
3266 volblocksize
)) != 0 ||
3267 (error
= zvol_check_volsize(volsize
,
3268 volblocksize
)) != 0)
3270 } else if (type
== DMU_OST_ZFS
) {
3274 * We have to have normalization and
3275 * case-folding flags correct when we do the
3276 * file system creation, so go figure them out
3279 VERIFY(nvlist_alloc(&zct
.zct_zplprops
,
3280 NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
3281 error
= zfs_fill_zplprops(fsname
, nvprops
,
3282 zct
.zct_zplprops
, &is_insensitive
);
3284 nvlist_free(zct
.zct_zplprops
);
3289 error
= dsl_crypto_params_create_nvlist(DCP_CMD_NONE
, nvprops
,
3292 nvlist_free(zct
.zct_zplprops
);
3296 error
= dmu_objset_create(fsname
, type
,
3297 is_insensitive
? DS_FLAG_CI_DATASET
: 0, dcp
, cbfunc
, &zct
);
3299 nvlist_free(zct
.zct_zplprops
);
3300 dsl_crypto_params_free(dcp
, !!error
);
3303 * It would be nice to do this atomically.
3306 error
= zfs_set_prop_nvlist(fsname
, ZPROP_SRC_LOCAL
,
3313 * Volumes will return EBUSY and cannot be destroyed
3314 * until all asynchronous minor handling has completed.
3315 * Wait for the spa_zvol_taskq to drain then retry.
3317 error2
= dsl_destroy_head(fsname
);
3318 while ((error2
== EBUSY
) && (type
== DMU_OST_ZVOL
)) {
3319 error2
= spa_open(fsname
, &spa
, FTAG
);
3321 taskq_wait(spa
->spa_zvol_taskq
);
3322 spa_close(spa
, FTAG
);
3324 error2
= dsl_destroy_head(fsname
);
3333 * "origin" -> name of origin snapshot
3334 * (optional) "props" -> { prop -> value }
3335 * (optional) "hidden_args" -> { "wkeydata" -> value }
3336 * raw uint8_t array of encryption wrapping key data (32 bytes)
3340 * outnvl: propname -> error code (int32)
3342 static const zfs_ioc_key_t zfs_keys_clone
[] = {
3343 {"origin", DATA_TYPE_STRING
, 0},
3344 {"props", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
3345 {"hidden_args", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
3349 zfs_ioc_clone(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3352 nvlist_t
*nvprops
= NULL
;
3355 origin_name
= fnvlist_lookup_string(innvl
, "origin");
3356 (void) nvlist_lookup_nvlist(innvl
, "props", &nvprops
);
3358 if (strchr(fsname
, '@') ||
3359 strchr(fsname
, '%'))
3360 return (SET_ERROR(EINVAL
));
3362 if (dataset_namecheck(origin_name
, NULL
, NULL
) != 0)
3363 return (SET_ERROR(EINVAL
));
3365 error
= dmu_objset_clone(fsname
, origin_name
);
3368 * It would be nice to do this atomically.
3371 error
= zfs_set_prop_nvlist(fsname
, ZPROP_SRC_LOCAL
,
3374 (void) dsl_destroy_head(fsname
);
3379 static const zfs_ioc_key_t zfs_keys_remap
[] = {
3385 zfs_ioc_remap(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3387 /* This IOCTL is no longer supported. */
3393 * "snaps" -> { snapshot1, snapshot2 }
3394 * (optional) "props" -> { prop -> value (string) }
3397 * outnvl: snapshot -> error code (int32)
3399 static const zfs_ioc_key_t zfs_keys_snapshot
[] = {
3400 {"snaps", DATA_TYPE_NVLIST
, 0},
3401 {"props", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
3405 zfs_ioc_snapshot(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3408 nvlist_t
*props
= NULL
;
3412 (void) nvlist_lookup_nvlist(innvl
, "props", &props
);
3413 if (!nvlist_empty(props
) &&
3414 zfs_earlier_version(poolname
, SPA_VERSION_SNAP_PROPS
))
3415 return (SET_ERROR(ENOTSUP
));
3416 if ((error
= zfs_check_userprops(props
)) != 0)
3419 snaps
= fnvlist_lookup_nvlist(innvl
, "snaps");
3420 poollen
= strlen(poolname
);
3421 for (pair
= nvlist_next_nvpair(snaps
, NULL
); pair
!= NULL
;
3422 pair
= nvlist_next_nvpair(snaps
, pair
)) {
3423 const char *name
= nvpair_name(pair
);
3424 char *cp
= strchr(name
, '@');
3427 * The snap name must contain an @, and the part after it must
3428 * contain only valid characters.
3431 zfs_component_namecheck(cp
+ 1, NULL
, NULL
) != 0)
3432 return (SET_ERROR(EINVAL
));
3435 * The snap must be in the specified pool.
3437 if (strncmp(name
, poolname
, poollen
) != 0 ||
3438 (name
[poollen
] != '/' && name
[poollen
] != '@'))
3439 return (SET_ERROR(EXDEV
));
3442 * Check for permission to set the properties on the fs.
3444 if (!nvlist_empty(props
)) {
3446 error
= zfs_secpolicy_write_perms(name
,
3447 ZFS_DELEG_PERM_USERPROP
, CRED());
3453 /* This must be the only snap of this fs. */
3454 for (nvpair_t
*pair2
= nvlist_next_nvpair(snaps
, pair
);
3455 pair2
!= NULL
; pair2
= nvlist_next_nvpair(snaps
, pair2
)) {
3456 if (strncmp(name
, nvpair_name(pair2
), cp
- name
+ 1)
3458 return (SET_ERROR(EXDEV
));
3463 error
= dsl_dataset_snapshot(snaps
, props
, outnvl
);
3469 * innvl: "message" -> string
3471 static const zfs_ioc_key_t zfs_keys_log_history
[] = {
3472 {"message", DATA_TYPE_STRING
, 0},
3477 zfs_ioc_log_history(const char *unused
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3485 * The poolname in the ioctl is not set, we get it from the TSD,
3486 * which was set at the end of the last successful ioctl that allows
3487 * logging. The secpolicy func already checked that it is set.
3488 * Only one log ioctl is allowed after each successful ioctl, so
3489 * we clear the TSD here.
3491 poolname
= tsd_get(zfs_allow_log_key
);
3492 if (poolname
== NULL
)
3493 return (SET_ERROR(EINVAL
));
3494 (void) tsd_set(zfs_allow_log_key
, NULL
);
3495 error
= spa_open(poolname
, &spa
, FTAG
);
3500 message
= fnvlist_lookup_string(innvl
, "message");
3502 if (spa_version(spa
) < SPA_VERSION_ZPOOL_HISTORY
) {
3503 spa_close(spa
, FTAG
);
3504 return (SET_ERROR(ENOTSUP
));
3507 error
= spa_history_log(spa
, message
);
3508 spa_close(spa
, FTAG
);
3513 * The dp_config_rwlock must not be held when calling this, because the
3514 * unmount may need to write out data.
3516 * This function is best-effort. Callers must deal gracefully if it
3517 * remains mounted (or is remounted after this call).
3519 * Returns 0 if the argument is not a snapshot, or it is not currently a
3520 * filesystem, or we were able to unmount it. Returns error code otherwise.
3523 zfs_unmount_snap(const char *snapname
)
3525 if (strchr(snapname
, '@') == NULL
)
3528 (void) zfsctl_snapshot_unmount((char *)snapname
, MNT_FORCE
);
3533 zfs_unmount_snap_cb(const char *snapname
, void *arg
)
3535 zfs_unmount_snap(snapname
);
3540 * When a clone is destroyed, its origin may also need to be destroyed,
3541 * in which case it must be unmounted. This routine will do that unmount
3545 zfs_destroy_unmount_origin(const char *fsname
)
3551 error
= dmu_objset_hold(fsname
, FTAG
, &os
);
3554 ds
= dmu_objset_ds(os
);
3555 if (dsl_dir_is_clone(ds
->ds_dir
) && DS_IS_DEFER_DESTROY(ds
->ds_prev
)) {
3556 char originname
[ZFS_MAX_DATASET_NAME_LEN
];
3557 dsl_dataset_name(ds
->ds_prev
, originname
);
3558 dmu_objset_rele(os
, FTAG
);
3559 zfs_unmount_snap(originname
);
3561 dmu_objset_rele(os
, FTAG
);
3567 * "snaps" -> { snapshot1, snapshot2 }
3568 * (optional boolean) "defer"
3571 * outnvl: snapshot -> error code (int32)
3573 static const zfs_ioc_key_t zfs_keys_destroy_snaps
[] = {
3574 {"snaps", DATA_TYPE_NVLIST
, 0},
3575 {"defer", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
3580 zfs_ioc_destroy_snaps(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3588 snaps
= fnvlist_lookup_nvlist(innvl
, "snaps");
3589 defer
= nvlist_exists(innvl
, "defer");
3591 poollen
= strlen(poolname
);
3592 for (pair
= nvlist_next_nvpair(snaps
, NULL
); pair
!= NULL
;
3593 pair
= nvlist_next_nvpair(snaps
, pair
)) {
3594 const char *name
= nvpair_name(pair
);
3597 * The snap must be in the specified pool to prevent the
3598 * invalid removal of zvol minors below.
3600 if (strncmp(name
, poolname
, poollen
) != 0 ||
3601 (name
[poollen
] != '/' && name
[poollen
] != '@'))
3602 return (SET_ERROR(EXDEV
));
3604 zfs_unmount_snap(nvpair_name(pair
));
3605 if (spa_open(name
, &spa
, FTAG
) == 0) {
3606 zvol_remove_minors(spa
, name
, B_TRUE
);
3607 spa_close(spa
, FTAG
);
3611 return (dsl_destroy_snapshots_nvl(snaps
, defer
, outnvl
));
3615 * Create bookmarks. Bookmark names are of the form <fs>#<bmark>.
3616 * All bookmarks must be in the same pool.
3619 * bookmark1 -> snapshot1, bookmark2 -> snapshot2
3622 * outnvl: bookmark -> error code (int32)
3625 static const zfs_ioc_key_t zfs_keys_bookmark
[] = {
3626 {"<bookmark>...", DATA_TYPE_STRING
, ZK_WILDCARDLIST
},
3631 zfs_ioc_bookmark(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3633 for (nvpair_t
*pair
= nvlist_next_nvpair(innvl
, NULL
);
3634 pair
!= NULL
; pair
= nvlist_next_nvpair(innvl
, pair
)) {
3638 * Verify the snapshot argument.
3640 if (nvpair_value_string(pair
, &snap_name
) != 0)
3641 return (SET_ERROR(EINVAL
));
3644 /* Verify that the keys (bookmarks) are unique */
3645 for (nvpair_t
*pair2
= nvlist_next_nvpair(innvl
, pair
);
3646 pair2
!= NULL
; pair2
= nvlist_next_nvpair(innvl
, pair2
)) {
3647 if (strcmp(nvpair_name(pair
), nvpair_name(pair2
)) == 0)
3648 return (SET_ERROR(EINVAL
));
3652 return (dsl_bookmark_create(innvl
, outnvl
));
3657 * property 1, property 2, ...
3661 * bookmark name 1 -> { property 1, property 2, ... },
3662 * bookmark name 2 -> { property 1, property 2, ... }
3666 static const zfs_ioc_key_t zfs_keys_get_bookmarks
[] = {
3667 {"<property>...", DATA_TYPE_BOOLEAN
, ZK_WILDCARDLIST
| ZK_OPTIONAL
},
3671 zfs_ioc_get_bookmarks(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3673 return (dsl_get_bookmarks(fsname
, innvl
, outnvl
));
3677 * innvl is not used.
3680 * property 1, property 2, ...
3684 static const zfs_ioc_key_t zfs_keys_get_bookmark_props
[] = {
3690 zfs_ioc_get_bookmark_props(const char *bookmark
, nvlist_t
*innvl
,
3693 char fsname
[ZFS_MAX_DATASET_NAME_LEN
];
3696 bmname
= strchr(bookmark
, '#');
3698 return (SET_ERROR(EINVAL
));
3701 (void) strlcpy(fsname
, bookmark
, sizeof (fsname
));
3702 *(strchr(fsname
, '#')) = '\0';
3704 return (dsl_get_bookmark_props(fsname
, bmname
, outnvl
));
3709 * bookmark name 1, bookmark name 2
3712 * outnvl: bookmark -> error code (int32)
3715 static const zfs_ioc_key_t zfs_keys_destroy_bookmarks
[] = {
3716 {"<bookmark>...", DATA_TYPE_BOOLEAN
, ZK_WILDCARDLIST
},
3720 zfs_ioc_destroy_bookmarks(const char *poolname
, nvlist_t
*innvl
,
3725 poollen
= strlen(poolname
);
3726 for (nvpair_t
*pair
= nvlist_next_nvpair(innvl
, NULL
);
3727 pair
!= NULL
; pair
= nvlist_next_nvpair(innvl
, pair
)) {
3728 const char *name
= nvpair_name(pair
);
3729 const char *cp
= strchr(name
, '#');
3732 * The bookmark name must contain an #, and the part after it
3733 * must contain only valid characters.
3736 zfs_component_namecheck(cp
+ 1, NULL
, NULL
) != 0)
3737 return (SET_ERROR(EINVAL
));
3740 * The bookmark must be in the specified pool.
3742 if (strncmp(name
, poolname
, poollen
) != 0 ||
3743 (name
[poollen
] != '/' && name
[poollen
] != '#'))
3744 return (SET_ERROR(EXDEV
));
3747 error
= dsl_bookmark_destroy(innvl
, outnvl
);
3751 static const zfs_ioc_key_t zfs_keys_channel_program
[] = {
3752 {"program", DATA_TYPE_STRING
, 0},
3753 {"arg", DATA_TYPE_ANY
, 0},
3754 {"sync", DATA_TYPE_BOOLEAN_VALUE
, ZK_OPTIONAL
},
3755 {"instrlimit", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
3756 {"memlimit", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
3760 zfs_ioc_channel_program(const char *poolname
, nvlist_t
*innvl
,
3764 uint64_t instrlimit
, memlimit
;
3765 boolean_t sync_flag
;
3766 nvpair_t
*nvarg
= NULL
;
3768 program
= fnvlist_lookup_string(innvl
, ZCP_ARG_PROGRAM
);
3769 if (0 != nvlist_lookup_boolean_value(innvl
, ZCP_ARG_SYNC
, &sync_flag
)) {
3772 if (0 != nvlist_lookup_uint64(innvl
, ZCP_ARG_INSTRLIMIT
, &instrlimit
)) {
3773 instrlimit
= ZCP_DEFAULT_INSTRLIMIT
;
3775 if (0 != nvlist_lookup_uint64(innvl
, ZCP_ARG_MEMLIMIT
, &memlimit
)) {
3776 memlimit
= ZCP_DEFAULT_MEMLIMIT
;
3778 nvarg
= fnvlist_lookup_nvpair(innvl
, ZCP_ARG_ARGLIST
);
3780 if (instrlimit
== 0 || instrlimit
> zfs_lua_max_instrlimit
)
3781 return (SET_ERROR(EINVAL
));
3782 if (memlimit
== 0 || memlimit
> zfs_lua_max_memlimit
)
3783 return (SET_ERROR(EINVAL
));
3785 return (zcp_eval(poolname
, program
, sync_flag
, instrlimit
, memlimit
,
3793 static const zfs_ioc_key_t zfs_keys_pool_checkpoint
[] = {
3799 zfs_ioc_pool_checkpoint(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3801 return (spa_checkpoint(poolname
));
3808 static const zfs_ioc_key_t zfs_keys_pool_discard_checkpoint
[] = {
3814 zfs_ioc_pool_discard_checkpoint(const char *poolname
, nvlist_t
*innvl
,
3817 return (spa_checkpoint_discard(poolname
));
3822 * zc_name name of dataset to destroy
3823 * zc_defer_destroy mark for deferred destroy
3828 zfs_ioc_destroy(zfs_cmd_t
*zc
)
3831 dmu_objset_type_t ost
;
3834 err
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
);
3837 ost
= dmu_objset_type(os
);
3838 dmu_objset_rele(os
, FTAG
);
3840 if (ost
== DMU_OST_ZFS
)
3841 zfs_unmount_snap(zc
->zc_name
);
3843 if (strchr(zc
->zc_name
, '@')) {
3844 err
= dsl_destroy_snapshot(zc
->zc_name
, zc
->zc_defer_destroy
);
3846 err
= dsl_destroy_head(zc
->zc_name
);
3847 if (err
== EEXIST
) {
3849 * It is possible that the given DS may have
3850 * hidden child (%recv) datasets - "leftovers"
3851 * resulting from the previously interrupted
3854 * 6 extra bytes for /%recv
3856 char namebuf
[ZFS_MAX_DATASET_NAME_LEN
+ 6];
3858 if (snprintf(namebuf
, sizeof (namebuf
), "%s/%s",
3859 zc
->zc_name
, recv_clone_name
) >=
3861 return (SET_ERROR(EINVAL
));
3864 * Try to remove the hidden child (%recv) and after
3865 * that try to remove the target dataset.
3866 * If the hidden child (%recv) does not exist
3867 * the original error (EEXIST) will be returned
3869 err
= dsl_destroy_head(namebuf
);
3871 err
= dsl_destroy_head(zc
->zc_name
);
3872 else if (err
== ENOENT
)
3873 err
= SET_ERROR(EEXIST
);
3882 * "initialize_command" -> POOL_INITIALIZE_{CANCEL|START|SUSPEND} (uint64)
3883 * "initialize_vdevs": { -> guids to initialize (nvlist)
3884 * "vdev_path_1": vdev_guid_1, (uint64),
3885 * "vdev_path_2": vdev_guid_2, (uint64),
3891 * "initialize_vdevs": { -> initialization errors (nvlist)
3892 * "vdev_path_1": errno, see function body for possible errnos (uint64)
3893 * "vdev_path_2": errno, ... (uint64)
3898 * EINVAL is returned for an unknown commands or if any of the provided vdev
3899 * guids have be specified with a type other than uint64.
3901 static const zfs_ioc_key_t zfs_keys_pool_initialize
[] = {
3902 {ZPOOL_INITIALIZE_COMMAND
, DATA_TYPE_UINT64
, 0},
3903 {ZPOOL_INITIALIZE_VDEVS
, DATA_TYPE_NVLIST
, 0}
3907 zfs_ioc_pool_initialize(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3910 if (nvlist_lookup_uint64(innvl
, ZPOOL_INITIALIZE_COMMAND
,
3912 return (SET_ERROR(EINVAL
));
3915 if (!(cmd_type
== POOL_INITIALIZE_CANCEL
||
3916 cmd_type
== POOL_INITIALIZE_START
||
3917 cmd_type
== POOL_INITIALIZE_SUSPEND
)) {
3918 return (SET_ERROR(EINVAL
));
3921 nvlist_t
*vdev_guids
;
3922 if (nvlist_lookup_nvlist(innvl
, ZPOOL_INITIALIZE_VDEVS
,
3923 &vdev_guids
) != 0) {
3924 return (SET_ERROR(EINVAL
));
3927 for (nvpair_t
*pair
= nvlist_next_nvpair(vdev_guids
, NULL
);
3928 pair
!= NULL
; pair
= nvlist_next_nvpair(vdev_guids
, pair
)) {
3930 if (nvpair_value_uint64(pair
, &vdev_guid
) != 0) {
3931 return (SET_ERROR(EINVAL
));
3936 int error
= spa_open(poolname
, &spa
, FTAG
);
3940 nvlist_t
*vdev_errlist
= fnvlist_alloc();
3941 int total_errors
= spa_vdev_initialize(spa
, vdev_guids
, cmd_type
,
3944 if (fnvlist_size(vdev_errlist
) > 0) {
3945 fnvlist_add_nvlist(outnvl
, ZPOOL_INITIALIZE_VDEVS
,
3948 fnvlist_free(vdev_errlist
);
3950 spa_close(spa
, FTAG
);
3951 return (total_errors
> 0 ? EINVAL
: 0);
3956 * "trim_command" -> POOL_TRIM_{CANCEL|START|SUSPEND} (uint64)
3957 * "trim_vdevs": { -> guids to TRIM (nvlist)
3958 * "vdev_path_1": vdev_guid_1, (uint64),
3959 * "vdev_path_2": vdev_guid_2, (uint64),
3962 * "trim_rate" -> Target TRIM rate in bytes/sec.
3963 * "trim_secure" -> Set to request a secure TRIM.
3967 * "trim_vdevs": { -> TRIM errors (nvlist)
3968 * "vdev_path_1": errno, see function body for possible errnos (uint64)
3969 * "vdev_path_2": errno, ... (uint64)
3974 * EINVAL is returned for an unknown commands or if any of the provided vdev
3975 * guids have be specified with a type other than uint64.
3977 static const zfs_ioc_key_t zfs_keys_pool_trim
[] = {
3978 {ZPOOL_TRIM_COMMAND
, DATA_TYPE_UINT64
, 0},
3979 {ZPOOL_TRIM_VDEVS
, DATA_TYPE_NVLIST
, 0},
3980 {ZPOOL_TRIM_RATE
, DATA_TYPE_UINT64
, ZK_OPTIONAL
},
3981 {ZPOOL_TRIM_SECURE
, DATA_TYPE_BOOLEAN_VALUE
, ZK_OPTIONAL
},
3985 zfs_ioc_pool_trim(const char *poolname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
3988 if (nvlist_lookup_uint64(innvl
, ZPOOL_TRIM_COMMAND
, &cmd_type
) != 0)
3989 return (SET_ERROR(EINVAL
));
3991 if (!(cmd_type
== POOL_TRIM_CANCEL
||
3992 cmd_type
== POOL_TRIM_START
||
3993 cmd_type
== POOL_TRIM_SUSPEND
)) {
3994 return (SET_ERROR(EINVAL
));
3997 nvlist_t
*vdev_guids
;
3998 if (nvlist_lookup_nvlist(innvl
, ZPOOL_TRIM_VDEVS
, &vdev_guids
) != 0)
3999 return (SET_ERROR(EINVAL
));
4001 for (nvpair_t
*pair
= nvlist_next_nvpair(vdev_guids
, NULL
);
4002 pair
!= NULL
; pair
= nvlist_next_nvpair(vdev_guids
, pair
)) {
4004 if (nvpair_value_uint64(pair
, &vdev_guid
) != 0) {
4005 return (SET_ERROR(EINVAL
));
4009 /* Optional, defaults to maximum rate when not provided */
4011 if (nvlist_lookup_uint64(innvl
, ZPOOL_TRIM_RATE
, &rate
) != 0)
4014 /* Optional, defaults to standard TRIM when not provided */
4016 if (nvlist_lookup_boolean_value(innvl
, ZPOOL_TRIM_SECURE
,
4022 int error
= spa_open(poolname
, &spa
, FTAG
);
4026 nvlist_t
*vdev_errlist
= fnvlist_alloc();
4027 int total_errors
= spa_vdev_trim(spa
, vdev_guids
, cmd_type
,
4028 rate
, !!zfs_trim_metaslab_skip
, secure
, vdev_errlist
);
4030 if (fnvlist_size(vdev_errlist
) > 0)
4031 fnvlist_add_nvlist(outnvl
, ZPOOL_TRIM_VDEVS
, vdev_errlist
);
4033 fnvlist_free(vdev_errlist
);
4035 spa_close(spa
, FTAG
);
4036 return (total_errors
> 0 ? EINVAL
: 0);
4040 * This ioctl waits for activity of a particular type to complete. If there is
4041 * no activity of that type in progress, it returns immediately, and the
4042 * returned value "waited" is false. If there is activity in progress, and no
4043 * tag is passed in, the ioctl blocks until all activity of that type is
4044 * complete, and then returns with "waited" set to true.
4046 * If a tag is provided, it identifies a particular instance of an activity to
4047 * wait for. Currently, this is only valid for use with 'initialize', because
4048 * that is the only activity for which there can be multiple instances running
4049 * concurrently. In the case of 'initialize', the tag corresponds to the guid of
4050 * the vdev on which to wait.
4052 * If a thread waiting in the ioctl receives a signal, the call will return
4053 * immediately, and the return value will be EINTR.
4056 * "wait_activity" -> int32_t
4057 * (optional) "wait_tag" -> uint64_t
4060 * outnvl: "waited" -> boolean_t
4062 static const zfs_ioc_key_t zfs_keys_pool_wait
[] = {
4063 {ZPOOL_WAIT_ACTIVITY
, DATA_TYPE_INT32
, 0},
4064 {ZPOOL_WAIT_TAG
, DATA_TYPE_UINT64
, ZK_OPTIONAL
},
4068 zfs_ioc_wait(const char *name
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
4075 if (nvlist_lookup_int32(innvl
, ZPOOL_WAIT_ACTIVITY
, &activity
) != 0)
4078 if (nvlist_lookup_uint64(innvl
, ZPOOL_WAIT_TAG
, &tag
) == 0)
4079 error
= spa_wait_tag(name
, activity
, tag
, &waited
);
4081 error
= spa_wait(name
, activity
, &waited
);
4084 fnvlist_add_boolean_value(outnvl
, ZPOOL_WAIT_WAITED
, waited
);
4090 * fsname is name of dataset to rollback (to most recent snapshot)
4092 * innvl may contain name of expected target snapshot
4094 * outnvl: "target" -> name of most recent snapshot
4097 static const zfs_ioc_key_t zfs_keys_rollback
[] = {
4098 {"target", DATA_TYPE_STRING
, ZK_OPTIONAL
},
4103 zfs_ioc_rollback(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
4106 zvol_state_handle_t
*zv
;
4107 char *target
= NULL
;
4110 (void) nvlist_lookup_string(innvl
, "target", &target
);
4111 if (target
!= NULL
) {
4112 const char *cp
= strchr(target
, '@');
4115 * The snap name must contain an @, and the part after it must
4116 * contain only valid characters.
4119 zfs_component_namecheck(cp
+ 1, NULL
, NULL
) != 0)
4120 return (SET_ERROR(EINVAL
));
4123 if (getzfsvfs(fsname
, &zfsvfs
) == 0) {
4126 ds
= dmu_objset_ds(zfsvfs
->z_os
);
4127 error
= zfs_suspend_fs(zfsvfs
);
4131 error
= dsl_dataset_rollback(fsname
, target
, zfsvfs
,
4133 resume_err
= zfs_resume_fs(zfsvfs
, ds
);
4134 error
= error
? error
: resume_err
;
4136 deactivate_super(zfsvfs
->z_sb
);
4137 } else if ((zv
= zvol_suspend(fsname
)) != NULL
) {
4138 error
= dsl_dataset_rollback(fsname
, target
, zvol_tag(zv
),
4142 error
= dsl_dataset_rollback(fsname
, target
, NULL
, outnvl
);
4148 recursive_unmount(const char *fsname
, void *arg
)
4150 const char *snapname
= arg
;
4153 fullname
= kmem_asprintf("%s@%s", fsname
, snapname
);
4154 zfs_unmount_snap(fullname
);
4162 * snapname is the snapshot to redact.
4164 * "bookname" -> (string)
4165 * name of the redaction bookmark to generate
4166 * "snapnv" -> (nvlist, values ignored)
4167 * snapshots to redact snapname with respect to
4174 static const zfs_ioc_key_t zfs_keys_redact
[] = {
4175 {"bookname", DATA_TYPE_STRING
, 0},
4176 {"snapnv", DATA_TYPE_NVLIST
, 0},
4179 zfs_ioc_redact(const char *snapname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
4181 nvlist_t
*redactnvl
= NULL
;
4182 char *redactbook
= NULL
;
4184 if (nvlist_lookup_nvlist(innvl
, "snapnv", &redactnvl
) != 0)
4185 return (SET_ERROR(EINVAL
));
4186 if (fnvlist_num_pairs(redactnvl
) == 0)
4187 return (SET_ERROR(ENXIO
));
4188 if (nvlist_lookup_string(innvl
, "bookname", &redactbook
) != 0)
4189 return (SET_ERROR(EINVAL
));
4191 return (dmu_redact_snap(snapname
, redactnvl
, redactbook
));
4196 * zc_name old name of dataset
4197 * zc_value new name of dataset
4198 * zc_cookie recursive flag (only valid for snapshots)
4203 zfs_ioc_rename(zfs_cmd_t
*zc
)
4206 dmu_objset_type_t ost
;
4207 boolean_t recursive
= zc
->zc_cookie
& 1;
4211 /* "zfs rename" from and to ...%recv datasets should both fail */
4212 zc
->zc_name
[sizeof (zc
->zc_name
) - 1] = '\0';
4213 zc
->zc_value
[sizeof (zc
->zc_value
) - 1] = '\0';
4214 if (dataset_namecheck(zc
->zc_name
, NULL
, NULL
) != 0 ||
4215 dataset_namecheck(zc
->zc_value
, NULL
, NULL
) != 0 ||
4216 strchr(zc
->zc_name
, '%') || strchr(zc
->zc_value
, '%'))
4217 return (SET_ERROR(EINVAL
));
4219 err
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
);
4222 ost
= dmu_objset_type(os
);
4223 dmu_objset_rele(os
, FTAG
);
4225 at
= strchr(zc
->zc_name
, '@');
4227 /* snaps must be in same fs */
4230 if (strncmp(zc
->zc_name
, zc
->zc_value
, at
- zc
->zc_name
+ 1))
4231 return (SET_ERROR(EXDEV
));
4233 if (ost
== DMU_OST_ZFS
) {
4234 error
= dmu_objset_find(zc
->zc_name
,
4235 recursive_unmount
, at
+ 1,
4236 recursive
? DS_FIND_CHILDREN
: 0);
4242 error
= dsl_dataset_rename_snapshot(zc
->zc_name
,
4243 at
+ 1, strchr(zc
->zc_value
, '@') + 1, recursive
);
4248 return (dsl_dir_rename(zc
->zc_name
, zc
->zc_value
));
4253 zfs_check_settable(const char *dsname
, nvpair_t
*pair
, cred_t
*cr
)
4255 const char *propname
= nvpair_name(pair
);
4256 boolean_t issnap
= (strchr(dsname
, '@') != NULL
);
4257 zfs_prop_t prop
= zfs_name_to_prop(propname
);
4261 if (prop
== ZPROP_INVAL
) {
4262 if (zfs_prop_user(propname
)) {
4263 if ((err
= zfs_secpolicy_write_perms(dsname
,
4264 ZFS_DELEG_PERM_USERPROP
, cr
)))
4269 if (!issnap
&& zfs_prop_userquota(propname
)) {
4270 const char *perm
= NULL
;
4271 const char *uq_prefix
=
4272 zfs_userquota_prop_prefixes
[ZFS_PROP_USERQUOTA
];
4273 const char *gq_prefix
=
4274 zfs_userquota_prop_prefixes
[ZFS_PROP_GROUPQUOTA
];
4275 const char *uiq_prefix
=
4276 zfs_userquota_prop_prefixes
[ZFS_PROP_USEROBJQUOTA
];
4277 const char *giq_prefix
=
4278 zfs_userquota_prop_prefixes
[ZFS_PROP_GROUPOBJQUOTA
];
4279 const char *pq_prefix
=
4280 zfs_userquota_prop_prefixes
[ZFS_PROP_PROJECTQUOTA
];
4281 const char *piq_prefix
= zfs_userquota_prop_prefixes
[\
4282 ZFS_PROP_PROJECTOBJQUOTA
];
4284 if (strncmp(propname
, uq_prefix
,
4285 strlen(uq_prefix
)) == 0) {
4286 perm
= ZFS_DELEG_PERM_USERQUOTA
;
4287 } else if (strncmp(propname
, uiq_prefix
,
4288 strlen(uiq_prefix
)) == 0) {
4289 perm
= ZFS_DELEG_PERM_USEROBJQUOTA
;
4290 } else if (strncmp(propname
, gq_prefix
,
4291 strlen(gq_prefix
)) == 0) {
4292 perm
= ZFS_DELEG_PERM_GROUPQUOTA
;
4293 } else if (strncmp(propname
, giq_prefix
,
4294 strlen(giq_prefix
)) == 0) {
4295 perm
= ZFS_DELEG_PERM_GROUPOBJQUOTA
;
4296 } else if (strncmp(propname
, pq_prefix
,
4297 strlen(pq_prefix
)) == 0) {
4298 perm
= ZFS_DELEG_PERM_PROJECTQUOTA
;
4299 } else if (strncmp(propname
, piq_prefix
,
4300 strlen(piq_prefix
)) == 0) {
4301 perm
= ZFS_DELEG_PERM_PROJECTOBJQUOTA
;
4303 /* {USER|GROUP|PROJECT}USED are read-only */
4304 return (SET_ERROR(EINVAL
));
4307 if ((err
= zfs_secpolicy_write_perms(dsname
, perm
, cr
)))
4312 return (SET_ERROR(EINVAL
));
4316 return (SET_ERROR(EINVAL
));
4318 if (nvpair_type(pair
) == DATA_TYPE_NVLIST
) {
4320 * dsl_prop_get_all_impl() returns properties in this
4324 VERIFY(nvpair_value_nvlist(pair
, &attrs
) == 0);
4325 VERIFY(nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
4330 * Check that this value is valid for this pool version
4333 case ZFS_PROP_COMPRESSION
:
4335 * If the user specified gzip compression, make sure
4336 * the SPA supports it. We ignore any errors here since
4337 * we'll catch them later.
4339 if (nvpair_value_uint64(pair
, &intval
) == 0) {
4340 if (intval
>= ZIO_COMPRESS_GZIP_1
&&
4341 intval
<= ZIO_COMPRESS_GZIP_9
&&
4342 zfs_earlier_version(dsname
,
4343 SPA_VERSION_GZIP_COMPRESSION
)) {
4344 return (SET_ERROR(ENOTSUP
));
4347 if (intval
== ZIO_COMPRESS_ZLE
&&
4348 zfs_earlier_version(dsname
,
4349 SPA_VERSION_ZLE_COMPRESSION
))
4350 return (SET_ERROR(ENOTSUP
));
4352 if (intval
== ZIO_COMPRESS_LZ4
) {
4355 if ((err
= spa_open(dsname
, &spa
, FTAG
)) != 0)
4358 if (!spa_feature_is_enabled(spa
,
4359 SPA_FEATURE_LZ4_COMPRESS
)) {
4360 spa_close(spa
, FTAG
);
4361 return (SET_ERROR(ENOTSUP
));
4363 spa_close(spa
, FTAG
);
4367 * If this is a bootable dataset then
4368 * verify that the compression algorithm
4369 * is supported for booting. We must return
4370 * something other than ENOTSUP since it
4371 * implies a downrev pool version.
4373 if (zfs_is_bootfs(dsname
) &&
4374 !BOOTFS_COMPRESS_VALID(intval
)) {
4375 return (SET_ERROR(ERANGE
));
4380 case ZFS_PROP_COPIES
:
4381 if (zfs_earlier_version(dsname
, SPA_VERSION_DITTO_BLOCKS
))
4382 return (SET_ERROR(ENOTSUP
));
4385 case ZFS_PROP_VOLBLOCKSIZE
:
4386 case ZFS_PROP_RECORDSIZE
:
4387 /* Record sizes above 128k need the feature to be enabled */
4388 if (nvpair_value_uint64(pair
, &intval
) == 0 &&
4389 intval
> SPA_OLD_MAXBLOCKSIZE
) {
4393 * We don't allow setting the property above 1MB,
4394 * unless the tunable has been changed.
4396 if (intval
> zfs_max_recordsize
||
4397 intval
> SPA_MAXBLOCKSIZE
)
4398 return (SET_ERROR(ERANGE
));
4400 if ((err
= spa_open(dsname
, &spa
, FTAG
)) != 0)
4403 if (!spa_feature_is_enabled(spa
,
4404 SPA_FEATURE_LARGE_BLOCKS
)) {
4405 spa_close(spa
, FTAG
);
4406 return (SET_ERROR(ENOTSUP
));
4408 spa_close(spa
, FTAG
);
4412 case ZFS_PROP_DNODESIZE
:
4413 /* Dnode sizes above 512 need the feature to be enabled */
4414 if (nvpair_value_uint64(pair
, &intval
) == 0 &&
4415 intval
!= ZFS_DNSIZE_LEGACY
) {
4419 * If this is a bootable dataset then
4420 * we don't allow large (>512B) dnodes,
4421 * because GRUB doesn't support them.
4423 if (zfs_is_bootfs(dsname
) &&
4424 intval
!= ZFS_DNSIZE_LEGACY
) {
4425 return (SET_ERROR(EDOM
));
4428 if ((err
= spa_open(dsname
, &spa
, FTAG
)) != 0)
4431 if (!spa_feature_is_enabled(spa
,
4432 SPA_FEATURE_LARGE_DNODE
)) {
4433 spa_close(spa
, FTAG
);
4434 return (SET_ERROR(ENOTSUP
));
4436 spa_close(spa
, FTAG
);
4440 case ZFS_PROP_SPECIAL_SMALL_BLOCKS
:
4442 * This property could require the allocation classes
4443 * feature to be active for setting, however we allow
4444 * it so that tests of settable properties succeed.
4445 * The CLI will issue a warning in this case.
4449 case ZFS_PROP_SHARESMB
:
4450 if (zpl_earlier_version(dsname
, ZPL_VERSION_FUID
))
4451 return (SET_ERROR(ENOTSUP
));
4454 case ZFS_PROP_ACLINHERIT
:
4455 if (nvpair_type(pair
) == DATA_TYPE_UINT64
&&
4456 nvpair_value_uint64(pair
, &intval
) == 0) {
4457 if (intval
== ZFS_ACL_PASSTHROUGH_X
&&
4458 zfs_earlier_version(dsname
,
4459 SPA_VERSION_PASSTHROUGH_X
))
4460 return (SET_ERROR(ENOTSUP
));
4463 case ZFS_PROP_CHECKSUM
:
4464 case ZFS_PROP_DEDUP
:
4466 spa_feature_t feature
;
4470 /* dedup feature version checks */
4471 if (prop
== ZFS_PROP_DEDUP
&&
4472 zfs_earlier_version(dsname
, SPA_VERSION_DEDUP
))
4473 return (SET_ERROR(ENOTSUP
));
4475 if (nvpair_type(pair
) == DATA_TYPE_UINT64
&&
4476 nvpair_value_uint64(pair
, &intval
) == 0) {
4477 /* check prop value is enabled in features */
4478 feature
= zio_checksum_to_feature(
4479 intval
& ZIO_CHECKSUM_MASK
);
4480 if (feature
== SPA_FEATURE_NONE
)
4483 if ((err
= spa_open(dsname
, &spa
, FTAG
)) != 0)
4486 if (!spa_feature_is_enabled(spa
, feature
)) {
4487 spa_close(spa
, FTAG
);
4488 return (SET_ERROR(ENOTSUP
));
4490 spa_close(spa
, FTAG
);
4499 return (zfs_secpolicy_setprop(dsname
, prop
, pair
, CRED()));
4503 * Removes properties from the given props list that fail permission checks
4504 * needed to clear them and to restore them in case of a receive error. For each
4505 * property, make sure we have both set and inherit permissions.
4507 * Returns the first error encountered if any permission checks fail. If the
4508 * caller provides a non-NULL errlist, it also gives the complete list of names
4509 * of all the properties that failed a permission check along with the
4510 * corresponding error numbers. The caller is responsible for freeing the
4513 * If every property checks out successfully, zero is returned and the list
4514 * pointed at by errlist is NULL.
4517 zfs_check_clearable(char *dataset
, nvlist_t
*props
, nvlist_t
**errlist
)
4520 nvpair_t
*pair
, *next_pair
;
4527 VERIFY(nvlist_alloc(&errors
, NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
4529 zc
= kmem_alloc(sizeof (zfs_cmd_t
), KM_SLEEP
);
4530 (void) strlcpy(zc
->zc_name
, dataset
, sizeof (zc
->zc_name
));
4531 pair
= nvlist_next_nvpair(props
, NULL
);
4532 while (pair
!= NULL
) {
4533 next_pair
= nvlist_next_nvpair(props
, pair
);
4535 (void) strlcpy(zc
->zc_value
, nvpair_name(pair
),
4536 sizeof (zc
->zc_value
));
4537 if ((err
= zfs_check_settable(dataset
, pair
, CRED())) != 0 ||
4538 (err
= zfs_secpolicy_inherit_prop(zc
, NULL
, CRED())) != 0) {
4539 VERIFY(nvlist_remove_nvpair(props
, pair
) == 0);
4540 VERIFY(nvlist_add_int32(errors
,
4541 zc
->zc_value
, err
) == 0);
4545 kmem_free(zc
, sizeof (zfs_cmd_t
));
4547 if ((pair
= nvlist_next_nvpair(errors
, NULL
)) == NULL
) {
4548 nvlist_free(errors
);
4551 VERIFY(nvpair_value_int32(pair
, &rv
) == 0);
4554 if (errlist
== NULL
)
4555 nvlist_free(errors
);
4563 propval_equals(nvpair_t
*p1
, nvpair_t
*p2
)
4565 if (nvpair_type(p1
) == DATA_TYPE_NVLIST
) {
4566 /* dsl_prop_get_all_impl() format */
4568 VERIFY(nvpair_value_nvlist(p1
, &attrs
) == 0);
4569 VERIFY(nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
4573 if (nvpair_type(p2
) == DATA_TYPE_NVLIST
) {
4575 VERIFY(nvpair_value_nvlist(p2
, &attrs
) == 0);
4576 VERIFY(nvlist_lookup_nvpair(attrs
, ZPROP_VALUE
,
4580 if (nvpair_type(p1
) != nvpair_type(p2
))
4583 if (nvpair_type(p1
) == DATA_TYPE_STRING
) {
4584 char *valstr1
, *valstr2
;
4586 VERIFY(nvpair_value_string(p1
, (char **)&valstr1
) == 0);
4587 VERIFY(nvpair_value_string(p2
, (char **)&valstr2
) == 0);
4588 return (strcmp(valstr1
, valstr2
) == 0);
4590 uint64_t intval1
, intval2
;
4592 VERIFY(nvpair_value_uint64(p1
, &intval1
) == 0);
4593 VERIFY(nvpair_value_uint64(p2
, &intval2
) == 0);
4594 return (intval1
== intval2
);
4599 * Remove properties from props if they are not going to change (as determined
4600 * by comparison with origprops). Remove them from origprops as well, since we
4601 * do not need to clear or restore properties that won't change.
4604 props_reduce(nvlist_t
*props
, nvlist_t
*origprops
)
4606 nvpair_t
*pair
, *next_pair
;
4608 if (origprops
== NULL
)
4609 return; /* all props need to be received */
4611 pair
= nvlist_next_nvpair(props
, NULL
);
4612 while (pair
!= NULL
) {
4613 const char *propname
= nvpair_name(pair
);
4616 next_pair
= nvlist_next_nvpair(props
, pair
);
4618 if ((nvlist_lookup_nvpair(origprops
, propname
,
4619 &match
) != 0) || !propval_equals(pair
, match
))
4620 goto next
; /* need to set received value */
4622 /* don't clear the existing received value */
4623 (void) nvlist_remove_nvpair(origprops
, match
);
4624 /* don't bother receiving the property */
4625 (void) nvlist_remove_nvpair(props
, pair
);
4632 * Extract properties that cannot be set PRIOR to the receipt of a dataset.
4633 * For example, refquota cannot be set until after the receipt of a dataset,
4634 * because in replication streams, an older/earlier snapshot may exceed the
4635 * refquota. We want to receive the older/earlier snapshot, but setting
4636 * refquota pre-receipt will set the dsl's ACTUAL quota, which will prevent
4637 * the older/earlier snapshot from being received (with EDQUOT).
4639 * The ZFS test "zfs_receive_011_pos" demonstrates such a scenario.
4641 * libzfs will need to be judicious handling errors encountered by props
4642 * extracted by this function.
4645 extract_delay_props(nvlist_t
*props
)
4647 nvlist_t
*delayprops
;
4648 nvpair_t
*nvp
, *tmp
;
4649 static const zfs_prop_t delayable
[] = {
4651 ZFS_PROP_KEYLOCATION
,
4656 VERIFY(nvlist_alloc(&delayprops
, NV_UNIQUE_NAME
, KM_SLEEP
) == 0);
4658 for (nvp
= nvlist_next_nvpair(props
, NULL
); nvp
!= NULL
;
4659 nvp
= nvlist_next_nvpair(props
, nvp
)) {
4661 * strcmp() is safe because zfs_prop_to_name() always returns
4664 for (i
= 0; delayable
[i
] != 0; i
++) {
4665 if (strcmp(zfs_prop_to_name(delayable
[i
]),
4666 nvpair_name(nvp
)) == 0) {
4670 if (delayable
[i
] != 0) {
4671 tmp
= nvlist_prev_nvpair(props
, nvp
);
4672 VERIFY(nvlist_add_nvpair(delayprops
, nvp
) == 0);
4673 VERIFY(nvlist_remove_nvpair(props
, nvp
) == 0);
4678 if (nvlist_empty(delayprops
)) {
4679 nvlist_free(delayprops
);
4682 return (delayprops
);
4686 zfs_allow_log_destroy(void *arg
)
4688 char *poolname
= arg
;
4690 if (poolname
!= NULL
)
4695 static boolean_t zfs_ioc_recv_inject_err
;
4699 * nvlist 'errors' is always allocated. It will contain descriptions of
4700 * encountered errors, if any. It's the callers responsibility to free.
4703 zfs_ioc_recv_impl(char *tofs
, char *tosnap
, char *origin
, nvlist_t
*recvprops
,
4704 nvlist_t
*localprops
, nvlist_t
*hidden_args
, boolean_t force
,
4705 boolean_t resumable
, int input_fd
, dmu_replay_record_t
*begin_record
,
4706 int cleanup_fd
, uint64_t *read_bytes
, uint64_t *errflags
,
4707 uint64_t *action_handle
, nvlist_t
**errors
)
4709 dmu_recv_cookie_t drc
;
4711 int props_error
= 0;
4713 nvlist_t
*local_delayprops
= NULL
;
4714 nvlist_t
*recv_delayprops
= NULL
;
4715 nvlist_t
*origprops
= NULL
; /* existing properties */
4716 nvlist_t
*origrecvd
= NULL
; /* existing received properties */
4717 boolean_t first_recvd_props
= B_FALSE
;
4718 boolean_t tofs_was_redacted
;
4723 *errors
= fnvlist_alloc();
4725 input_fp
= getf(input_fd
);
4726 if (input_fp
== NULL
)
4727 return (SET_ERROR(EBADF
));
4729 off
= input_fp
->f_offset
;
4730 error
= dmu_recv_begin(tofs
, tosnap
, begin_record
, force
,
4731 resumable
, localprops
, hidden_args
, origin
, &drc
, input_fp
->f_vnode
,
4735 tofs_was_redacted
= dsl_get_redacted(drc
.drc_ds
);
4738 * Set properties before we receive the stream so that they are applied
4739 * to the new data. Note that we must call dmu_recv_stream() if
4740 * dmu_recv_begin() succeeds.
4742 if (recvprops
!= NULL
&& !drc
.drc_newfs
) {
4743 if (spa_version(dsl_dataset_get_spa(drc
.drc_ds
)) >=
4744 SPA_VERSION_RECVD_PROPS
&&
4745 !dsl_prop_get_hasrecvd(tofs
))
4746 first_recvd_props
= B_TRUE
;
4749 * If new received properties are supplied, they are to
4750 * completely replace the existing received properties,
4751 * so stash away the existing ones.
4753 if (dsl_prop_get_received(tofs
, &origrecvd
) == 0) {
4754 nvlist_t
*errlist
= NULL
;
4756 * Don't bother writing a property if its value won't
4757 * change (and avoid the unnecessary security checks).
4759 * The first receive after SPA_VERSION_RECVD_PROPS is a
4760 * special case where we blow away all local properties
4763 if (!first_recvd_props
)
4764 props_reduce(recvprops
, origrecvd
);
4765 if (zfs_check_clearable(tofs
, origrecvd
, &errlist
) != 0)
4766 (void) nvlist_merge(*errors
, errlist
, 0);
4767 nvlist_free(errlist
);
4769 if (clear_received_props(tofs
, origrecvd
,
4770 first_recvd_props
? NULL
: recvprops
) != 0)
4771 *errflags
|= ZPROP_ERR_NOCLEAR
;
4773 *errflags
|= ZPROP_ERR_NOCLEAR
;
4778 * Stash away existing properties so we can restore them on error unless
4779 * we're doing the first receive after SPA_VERSION_RECVD_PROPS, in which
4780 * case "origrecvd" will take care of that.
4782 if (localprops
!= NULL
&& !drc
.drc_newfs
&& !first_recvd_props
) {
4784 if (dmu_objset_hold(tofs
, FTAG
, &os
) == 0) {
4785 if (dsl_prop_get_all(os
, &origprops
) != 0) {
4786 *errflags
|= ZPROP_ERR_NOCLEAR
;
4788 dmu_objset_rele(os
, FTAG
);
4790 *errflags
|= ZPROP_ERR_NOCLEAR
;
4794 if (recvprops
!= NULL
) {
4795 props_error
= dsl_prop_set_hasrecvd(tofs
);
4797 if (props_error
== 0) {
4798 recv_delayprops
= extract_delay_props(recvprops
);
4799 (void) zfs_set_prop_nvlist(tofs
, ZPROP_SRC_RECEIVED
,
4800 recvprops
, *errors
);
4804 if (localprops
!= NULL
) {
4805 nvlist_t
*oprops
= fnvlist_alloc();
4806 nvlist_t
*xprops
= fnvlist_alloc();
4807 nvpair_t
*nvp
= NULL
;
4809 while ((nvp
= nvlist_next_nvpair(localprops
, nvp
)) != NULL
) {
4810 if (nvpair_type(nvp
) == DATA_TYPE_BOOLEAN
) {
4812 const char *name
= nvpair_name(nvp
);
4813 zfs_prop_t prop
= zfs_name_to_prop(name
);
4814 if (prop
!= ZPROP_INVAL
) {
4815 if (!zfs_prop_inheritable(prop
))
4817 } else if (!zfs_prop_user(name
))
4819 fnvlist_add_boolean(xprops
, name
);
4821 /* -o property=value */
4822 fnvlist_add_nvpair(oprops
, nvp
);
4826 local_delayprops
= extract_delay_props(oprops
);
4827 (void) zfs_set_prop_nvlist(tofs
, ZPROP_SRC_LOCAL
,
4829 (void) zfs_set_prop_nvlist(tofs
, ZPROP_SRC_INHERITED
,
4832 nvlist_free(oprops
);
4833 nvlist_free(xprops
);
4836 error
= dmu_recv_stream(&drc
, cleanup_fd
, action_handle
, &off
);
4839 zfsvfs_t
*zfsvfs
= NULL
;
4840 zvol_state_handle_t
*zv
= NULL
;
4842 if (getzfsvfs(tofs
, &zfsvfs
) == 0) {
4846 boolean_t stream_is_redacted
= DMU_GET_FEATUREFLAGS(
4847 begin_record
->drr_u
.drr_begin
.
4848 drr_versioninfo
) & DMU_BACKUP_FEATURE_REDACTED
;
4850 ds
= dmu_objset_ds(zfsvfs
->z_os
);
4851 error
= zfs_suspend_fs(zfsvfs
);
4853 * If the suspend fails, then the recv_end will
4854 * likely also fail, and clean up after itself.
4856 end_err
= dmu_recv_end(&drc
, zfsvfs
);
4858 * If the dataset was not redacted, but we received a
4859 * redacted stream onto it, we need to unmount the
4860 * dataset. Otherwise, resume the filesystem.
4862 if (error
== 0 && !drc
.drc_newfs
&&
4863 stream_is_redacted
&& !tofs_was_redacted
) {
4864 error
= zfs_end_fs(zfsvfs
, ds
);
4865 } else if (error
== 0) {
4866 error
= zfs_resume_fs(zfsvfs
, ds
);
4868 error
= error
? error
: end_err
;
4869 deactivate_super(zfsvfs
->z_sb
);
4870 } else if ((zv
= zvol_suspend(tofs
)) != NULL
) {
4871 error
= dmu_recv_end(&drc
, zvol_tag(zv
));
4874 error
= dmu_recv_end(&drc
, NULL
);
4877 /* Set delayed properties now, after we're done receiving. */
4878 if (recv_delayprops
!= NULL
&& error
== 0) {
4879 (void) zfs_set_prop_nvlist(tofs
, ZPROP_SRC_RECEIVED
,
4880 recv_delayprops
, *errors
);
4882 if (local_delayprops
!= NULL
&& error
== 0) {
4883 (void) zfs_set_prop_nvlist(tofs
, ZPROP_SRC_LOCAL
,
4884 local_delayprops
, *errors
);
4889 * Merge delayed props back in with initial props, in case
4890 * we're DEBUG and zfs_ioc_recv_inject_err is set (which means
4891 * we have to make sure clear_received_props() includes
4892 * the delayed properties).
4894 * Since zfs_ioc_recv_inject_err is only in DEBUG kernels,
4895 * using ASSERT() will be just like a VERIFY.
4897 if (recv_delayprops
!= NULL
) {
4898 ASSERT(nvlist_merge(recvprops
, recv_delayprops
, 0) == 0);
4899 nvlist_free(recv_delayprops
);
4901 if (local_delayprops
!= NULL
) {
4902 ASSERT(nvlist_merge(localprops
, local_delayprops
, 0) == 0);
4903 nvlist_free(local_delayprops
);
4906 *read_bytes
= off
- input_fp
->f_offset
;
4907 if (VOP_SEEK(input_fp
->f_vnode
, input_fp
->f_offset
, &off
, NULL
) == 0)
4908 input_fp
->f_offset
= off
;
4911 if (zfs_ioc_recv_inject_err
) {
4912 zfs_ioc_recv_inject_err
= B_FALSE
;
4918 * On error, restore the original props.
4920 if (error
!= 0 && recvprops
!= NULL
&& !drc
.drc_newfs
) {
4921 if (clear_received_props(tofs
, recvprops
, NULL
) != 0) {
4923 * We failed to clear the received properties.
4924 * Since we may have left a $recvd value on the
4925 * system, we can't clear the $hasrecvd flag.
4927 *errflags
|= ZPROP_ERR_NORESTORE
;
4928 } else if (first_recvd_props
) {
4929 dsl_prop_unset_hasrecvd(tofs
);
4932 if (origrecvd
== NULL
&& !drc
.drc_newfs
) {
4933 /* We failed to stash the original properties. */
4934 *errflags
|= ZPROP_ERR_NORESTORE
;
4938 * dsl_props_set() will not convert RECEIVED to LOCAL on or
4939 * after SPA_VERSION_RECVD_PROPS, so we need to specify LOCAL
4940 * explicitly if we're restoring local properties cleared in the
4941 * first new-style receive.
4943 if (origrecvd
!= NULL
&&
4944 zfs_set_prop_nvlist(tofs
, (first_recvd_props
?
4945 ZPROP_SRC_LOCAL
: ZPROP_SRC_RECEIVED
),
4946 origrecvd
, NULL
) != 0) {
4948 * We stashed the original properties but failed to
4951 *errflags
|= ZPROP_ERR_NORESTORE
;
4954 if (error
!= 0 && localprops
!= NULL
&& !drc
.drc_newfs
&&
4955 !first_recvd_props
) {
4957 nvlist_t
*inheritprops
;
4960 if (origprops
== NULL
) {
4961 /* We failed to stash the original properties. */
4962 *errflags
|= ZPROP_ERR_NORESTORE
;
4966 /* Restore original props */
4967 setprops
= fnvlist_alloc();
4968 inheritprops
= fnvlist_alloc();
4970 while ((nvp
= nvlist_next_nvpair(localprops
, nvp
)) != NULL
) {
4971 const char *name
= nvpair_name(nvp
);
4975 if (!nvlist_exists(origprops
, name
)) {
4977 * Property was not present or was explicitly
4978 * inherited before the receive, restore this.
4980 fnvlist_add_boolean(inheritprops
, name
);
4983 attrs
= fnvlist_lookup_nvlist(origprops
, name
);
4984 source
= fnvlist_lookup_string(attrs
, ZPROP_SOURCE
);
4986 /* Skip received properties */
4987 if (strcmp(source
, ZPROP_SOURCE_VAL_RECVD
) == 0)
4990 if (strcmp(source
, tofs
) == 0) {
4991 /* Property was locally set */
4992 fnvlist_add_nvlist(setprops
, name
, attrs
);
4994 /* Property was implicitly inherited */
4995 fnvlist_add_boolean(inheritprops
, name
);
4999 if (zfs_set_prop_nvlist(tofs
, ZPROP_SRC_LOCAL
, setprops
,
5001 *errflags
|= ZPROP_ERR_NORESTORE
;
5002 if (zfs_set_prop_nvlist(tofs
, ZPROP_SRC_INHERITED
, inheritprops
,
5004 *errflags
|= ZPROP_ERR_NORESTORE
;
5006 nvlist_free(setprops
);
5007 nvlist_free(inheritprops
);
5011 nvlist_free(origrecvd
);
5012 nvlist_free(origprops
);
5015 error
= props_error
;
5022 * zc_name name of containing filesystem (unused)
5023 * zc_nvlist_src{_size} nvlist of properties to apply
5024 * zc_nvlist_conf{_size} nvlist of properties to exclude
5025 * (DATA_TYPE_BOOLEAN) and override (everything else)
5026 * zc_value name of snapshot to create
5027 * zc_string name of clone origin (if DRR_FLAG_CLONE)
5028 * zc_cookie file descriptor to recv from
5029 * zc_begin_record the BEGIN record of the stream (not byteswapped)
5030 * zc_guid force flag
5031 * zc_cleanup_fd cleanup-on-exit file descriptor
5032 * zc_action_handle handle for this guid/ds mapping (or zero on first call)
5035 * zc_cookie number of bytes read
5036 * zc_obj zprop_errflags_t
5037 * zc_action_handle handle for this guid/ds mapping
5038 * zc_nvlist_dst{_size} error for each unapplied received property
5041 zfs_ioc_recv(zfs_cmd_t
*zc
)
5043 dmu_replay_record_t begin_record
;
5044 nvlist_t
*errors
= NULL
;
5045 nvlist_t
*recvdprops
= NULL
;
5046 nvlist_t
*localprops
= NULL
;
5047 char *origin
= NULL
;
5049 char tofs
[ZFS_MAX_DATASET_NAME_LEN
];
5052 if (dataset_namecheck(zc
->zc_value
, NULL
, NULL
) != 0 ||
5053 strchr(zc
->zc_value
, '@') == NULL
||
5054 strchr(zc
->zc_value
, '%'))
5055 return (SET_ERROR(EINVAL
));
5057 (void) strlcpy(tofs
, zc
->zc_value
, sizeof (tofs
));
5058 tosnap
= strchr(tofs
, '@');
5061 if (zc
->zc_nvlist_src
!= 0 &&
5062 (error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
5063 zc
->zc_iflags
, &recvdprops
)) != 0)
5066 if (zc
->zc_nvlist_conf
!= 0 &&
5067 (error
= get_nvlist(zc
->zc_nvlist_conf
, zc
->zc_nvlist_conf_size
,
5068 zc
->zc_iflags
, &localprops
)) != 0)
5071 if (zc
->zc_string
[0])
5072 origin
= zc
->zc_string
;
5074 begin_record
.drr_type
= DRR_BEGIN
;
5075 begin_record
.drr_payloadlen
= 0;
5076 begin_record
.drr_u
.drr_begin
= zc
->zc_begin_record
;
5078 error
= zfs_ioc_recv_impl(tofs
, tosnap
, origin
, recvdprops
, localprops
,
5079 NULL
, zc
->zc_guid
, B_FALSE
, zc
->zc_cookie
, &begin_record
,
5080 zc
->zc_cleanup_fd
, &zc
->zc_cookie
, &zc
->zc_obj
,
5081 &zc
->zc_action_handle
, &errors
);
5082 nvlist_free(recvdprops
);
5083 nvlist_free(localprops
);
5086 * Now that all props, initial and delayed, are set, report the prop
5087 * errors to the caller.
5089 if (zc
->zc_nvlist_dst_size
!= 0 && errors
!= NULL
&&
5090 (nvlist_smush(errors
, zc
->zc_nvlist_dst_size
) != 0 ||
5091 put_nvlist(zc
, errors
) != 0)) {
5093 * Caller made zc->zc_nvlist_dst less than the minimum expected
5094 * size or supplied an invalid address.
5096 error
= SET_ERROR(EINVAL
);
5099 nvlist_free(errors
);
5106 * "snapname" -> full name of the snapshot to create
5107 * (optional) "props" -> received properties to set (nvlist)
5108 * (optional) "localprops" -> override and exclude properties (nvlist)
5109 * (optional) "origin" -> name of clone origin (DRR_FLAG_CLONE)
5110 * "begin_record" -> non-byteswapped dmu_replay_record_t
5111 * "input_fd" -> file descriptor to read stream from (int32)
5112 * (optional) "force" -> force flag (value ignored)
5113 * (optional) "resumable" -> resumable flag (value ignored)
5114 * (optional) "cleanup_fd" -> cleanup-on-exit file descriptor
5115 * (optional) "action_handle" -> handle for this guid/ds mapping
5116 * (optional) "hidden_args" -> { "wkeydata" -> value }
5120 * "read_bytes" -> number of bytes read
5121 * "error_flags" -> zprop_errflags_t
5122 * "action_handle" -> handle for this guid/ds mapping
5123 * "errors" -> error for each unapplied received property (nvlist)
5126 static const zfs_ioc_key_t zfs_keys_recv_new
[] = {
5127 {"snapname", DATA_TYPE_STRING
, 0},
5128 {"props", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
5129 {"localprops", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
5130 {"origin", DATA_TYPE_STRING
, ZK_OPTIONAL
},
5131 {"begin_record", DATA_TYPE_BYTE_ARRAY
, 0},
5132 {"input_fd", DATA_TYPE_INT32
, 0},
5133 {"force", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
5134 {"resumable", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
5135 {"cleanup_fd", DATA_TYPE_INT32
, ZK_OPTIONAL
},
5136 {"action_handle", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
5137 {"hidden_args", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
5141 zfs_ioc_recv_new(const char *fsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
5143 dmu_replay_record_t
*begin_record
;
5144 uint_t begin_record_size
;
5145 nvlist_t
*errors
= NULL
;
5146 nvlist_t
*recvprops
= NULL
;
5147 nvlist_t
*localprops
= NULL
;
5148 nvlist_t
*hidden_args
= NULL
;
5150 char *origin
= NULL
;
5152 char tofs
[ZFS_MAX_DATASET_NAME_LEN
];
5154 boolean_t resumable
;
5155 uint64_t action_handle
= 0;
5156 uint64_t read_bytes
= 0;
5157 uint64_t errflags
= 0;
5159 int cleanup_fd
= -1;
5162 snapname
= fnvlist_lookup_string(innvl
, "snapname");
5164 if (dataset_namecheck(snapname
, NULL
, NULL
) != 0 ||
5165 strchr(snapname
, '@') == NULL
||
5166 strchr(snapname
, '%'))
5167 return (SET_ERROR(EINVAL
));
5169 (void) strcpy(tofs
, snapname
);
5170 tosnap
= strchr(tofs
, '@');
5173 error
= nvlist_lookup_string(innvl
, "origin", &origin
);
5174 if (error
&& error
!= ENOENT
)
5177 error
= nvlist_lookup_byte_array(innvl
, "begin_record",
5178 (uchar_t
**)&begin_record
, &begin_record_size
);
5179 if (error
!= 0 || begin_record_size
!= sizeof (*begin_record
))
5180 return (SET_ERROR(EINVAL
));
5182 input_fd
= fnvlist_lookup_int32(innvl
, "input_fd");
5184 force
= nvlist_exists(innvl
, "force");
5185 resumable
= nvlist_exists(innvl
, "resumable");
5187 error
= nvlist_lookup_int32(innvl
, "cleanup_fd", &cleanup_fd
);
5188 if (error
&& error
!= ENOENT
)
5191 error
= nvlist_lookup_uint64(innvl
, "action_handle", &action_handle
);
5192 if (error
&& error
!= ENOENT
)
5195 /* we still use "props" here for backwards compatibility */
5196 error
= nvlist_lookup_nvlist(innvl
, "props", &recvprops
);
5197 if (error
&& error
!= ENOENT
)
5200 error
= nvlist_lookup_nvlist(innvl
, "localprops", &localprops
);
5201 if (error
&& error
!= ENOENT
)
5204 error
= nvlist_lookup_nvlist(innvl
, ZPOOL_HIDDEN_ARGS
, &hidden_args
);
5205 if (error
&& error
!= ENOENT
)
5208 error
= zfs_ioc_recv_impl(tofs
, tosnap
, origin
, recvprops
, localprops
,
5209 hidden_args
, force
, resumable
, input_fd
, begin_record
, cleanup_fd
,
5210 &read_bytes
, &errflags
, &action_handle
, &errors
);
5212 fnvlist_add_uint64(outnvl
, "read_bytes", read_bytes
);
5213 fnvlist_add_uint64(outnvl
, "error_flags", errflags
);
5214 fnvlist_add_uint64(outnvl
, "action_handle", action_handle
);
5215 fnvlist_add_nvlist(outnvl
, "errors", errors
);
5217 nvlist_free(errors
);
5218 nvlist_free(recvprops
);
5219 nvlist_free(localprops
);
5224 typedef struct dump_bytes_io
{
5232 dump_bytes_cb(void *arg
)
5234 dump_bytes_io_t
*dbi
= (dump_bytes_io_t
*)arg
;
5235 ssize_t resid
; /* have to get resid to get detailed errno */
5237 dbi
->dbi_err
= vn_rdwr(UIO_WRITE
, dbi
->dbi_vp
,
5238 (caddr_t
)dbi
->dbi_buf
, dbi
->dbi_len
,
5239 0, UIO_SYSSPACE
, FAPPEND
, RLIM64_INFINITY
, CRED(), &resid
);
5243 dump_bytes(objset_t
*os
, void *buf
, int len
, void *arg
)
5245 dump_bytes_io_t dbi
;
5251 #if defined(HAVE_LARGE_STACKS)
5252 dump_bytes_cb(&dbi
);
5255 * The vn_rdwr() call is performed in a taskq to ensure that there is
5256 * always enough stack space to write safely to the target filesystem.
5257 * The ZIO_TYPE_FREE threads are used because there can be a lot of
5258 * them and they are used in vdev_file.c for a similar purpose.
5260 spa_taskq_dispatch_sync(dmu_objset_spa(os
), ZIO_TYPE_FREE
,
5261 ZIO_TASKQ_ISSUE
, dump_bytes_cb
, &dbi
, TQ_SLEEP
);
5262 #endif /* HAVE_LARGE_STACKS */
5264 return (dbi
.dbi_err
);
5269 * zc_name name of snapshot to send
5270 * zc_cookie file descriptor to send stream to
5271 * zc_obj fromorigin flag (mutually exclusive with zc_fromobj)
5272 * zc_sendobj objsetid of snapshot to send
5273 * zc_fromobj objsetid of incremental fromsnap (may be zero)
5274 * zc_guid if set, estimate size of stream only. zc_cookie is ignored.
5275 * output size in zc_objset_type.
5276 * zc_flags lzc_send_flags
5279 * zc_objset_type estimated size, if zc_guid is set
5281 * NOTE: This is no longer the preferred interface, any new functionality
5282 * should be added to zfs_ioc_send_new() instead.
5285 zfs_ioc_send(zfs_cmd_t
*zc
)
5289 boolean_t estimate
= (zc
->zc_guid
!= 0);
5290 boolean_t embedok
= (zc
->zc_flags
& 0x1);
5291 boolean_t large_block_ok
= (zc
->zc_flags
& 0x2);
5292 boolean_t compressok
= (zc
->zc_flags
& 0x4);
5293 boolean_t rawok
= (zc
->zc_flags
& 0x8);
5295 if (zc
->zc_obj
!= 0) {
5297 dsl_dataset_t
*tosnap
;
5299 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
5303 error
= dsl_dataset_hold_obj(dp
, zc
->zc_sendobj
, FTAG
, &tosnap
);
5305 dsl_pool_rele(dp
, FTAG
);
5309 if (dsl_dir_is_clone(tosnap
->ds_dir
))
5311 dsl_dir_phys(tosnap
->ds_dir
)->dd_origin_obj
;
5312 dsl_dataset_rele(tosnap
, FTAG
);
5313 dsl_pool_rele(dp
, FTAG
);
5318 dsl_dataset_t
*tosnap
;
5319 dsl_dataset_t
*fromsnap
= NULL
;
5321 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
5325 error
= dsl_dataset_hold_obj(dp
, zc
->zc_sendobj
,
5328 dsl_pool_rele(dp
, FTAG
);
5332 if (zc
->zc_fromobj
!= 0) {
5333 error
= dsl_dataset_hold_obj(dp
, zc
->zc_fromobj
,
5336 dsl_dataset_rele(tosnap
, FTAG
);
5337 dsl_pool_rele(dp
, FTAG
);
5342 error
= dmu_send_estimate_fast(tosnap
, fromsnap
, NULL
,
5343 compressok
|| rawok
, &zc
->zc_objset_type
);
5345 if (fromsnap
!= NULL
)
5346 dsl_dataset_rele(fromsnap
, FTAG
);
5347 dsl_dataset_rele(tosnap
, FTAG
);
5348 dsl_pool_rele(dp
, FTAG
);
5350 file_t
*fp
= getf(zc
->zc_cookie
);
5352 return (SET_ERROR(EBADF
));
5355 dmu_send_outparams_t out
= {0};
5356 out
.dso_outfunc
= dump_bytes
;
5357 out
.dso_arg
= fp
->f_vnode
;
5358 out
.dso_dryrun
= B_FALSE
;
5359 error
= dmu_send_obj(zc
->zc_name
, zc
->zc_sendobj
,
5360 zc
->zc_fromobj
, embedok
, large_block_ok
, compressok
, rawok
,
5361 zc
->zc_cookie
, &off
, &out
);
5363 if (VOP_SEEK(fp
->f_vnode
, fp
->f_offset
, &off
, NULL
) == 0)
5365 releasef(zc
->zc_cookie
);
5372 * zc_name name of snapshot on which to report progress
5373 * zc_cookie file descriptor of send stream
5376 * zc_cookie number of bytes written in send stream thus far
5377 * zc_objset_type logical size of data traversed by send thus far
5380 zfs_ioc_send_progress(zfs_cmd_t
*zc
)
5384 dmu_sendstatus_t
*dsp
= NULL
;
5387 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
5391 error
= dsl_dataset_hold(dp
, zc
->zc_name
, FTAG
, &ds
);
5393 dsl_pool_rele(dp
, FTAG
);
5397 mutex_enter(&ds
->ds_sendstream_lock
);
5400 * Iterate over all the send streams currently active on this dataset.
5401 * If there's one which matches the specified file descriptor _and_ the
5402 * stream was started by the current process, return the progress of
5406 for (dsp
= list_head(&ds
->ds_sendstreams
); dsp
!= NULL
;
5407 dsp
= list_next(&ds
->ds_sendstreams
, dsp
)) {
5408 if (dsp
->dss_outfd
== zc
->zc_cookie
&&
5409 dsp
->dss_proc
== curproc
)
5414 zc
->zc_cookie
= atomic_cas_64((volatile uint64_t *)dsp
->dss_off
,
5416 /* This is the closest thing we have to atomic_read_64. */
5417 zc
->zc_objset_type
= atomic_cas_64(&dsp
->dss_blocks
, 0, 0);
5419 error
= SET_ERROR(ENOENT
);
5422 mutex_exit(&ds
->ds_sendstream_lock
);
5423 dsl_dataset_rele(ds
, FTAG
);
5424 dsl_pool_rele(dp
, FTAG
);
5429 zfs_ioc_inject_fault(zfs_cmd_t
*zc
)
5433 error
= zio_inject_fault(zc
->zc_name
, (int)zc
->zc_guid
, &id
,
5434 &zc
->zc_inject_record
);
5437 zc
->zc_guid
= (uint64_t)id
;
5443 zfs_ioc_clear_fault(zfs_cmd_t
*zc
)
5445 return (zio_clear_fault((int)zc
->zc_guid
));
5449 zfs_ioc_inject_list_next(zfs_cmd_t
*zc
)
5451 int id
= (int)zc
->zc_guid
;
5454 error
= zio_inject_list_next(&id
, zc
->zc_name
, sizeof (zc
->zc_name
),
5455 &zc
->zc_inject_record
);
5463 zfs_ioc_error_log(zfs_cmd_t
*zc
)
5467 size_t count
= (size_t)zc
->zc_nvlist_dst_size
;
5469 if ((error
= spa_open(zc
->zc_name
, &spa
, FTAG
)) != 0)
5472 error
= spa_get_errlog(spa
, (void *)(uintptr_t)zc
->zc_nvlist_dst
,
5475 zc
->zc_nvlist_dst_size
= count
;
5477 zc
->zc_nvlist_dst_size
= spa_get_errlog_size(spa
);
5479 spa_close(spa
, FTAG
);
5485 zfs_ioc_clear(zfs_cmd_t
*zc
)
5492 * On zpool clear we also fix up missing slogs
5494 mutex_enter(&spa_namespace_lock
);
5495 spa
= spa_lookup(zc
->zc_name
);
5497 mutex_exit(&spa_namespace_lock
);
5498 return (SET_ERROR(EIO
));
5500 if (spa_get_log_state(spa
) == SPA_LOG_MISSING
) {
5501 /* we need to let spa_open/spa_load clear the chains */
5502 spa_set_log_state(spa
, SPA_LOG_CLEAR
);
5504 spa
->spa_last_open_failed
= 0;
5505 mutex_exit(&spa_namespace_lock
);
5507 if (zc
->zc_cookie
& ZPOOL_NO_REWIND
) {
5508 error
= spa_open(zc
->zc_name
, &spa
, FTAG
);
5511 nvlist_t
*config
= NULL
;
5513 if (zc
->zc_nvlist_src
== 0)
5514 return (SET_ERROR(EINVAL
));
5516 if ((error
= get_nvlist(zc
->zc_nvlist_src
,
5517 zc
->zc_nvlist_src_size
, zc
->zc_iflags
, &policy
)) == 0) {
5518 error
= spa_open_rewind(zc
->zc_name
, &spa
, FTAG
,
5520 if (config
!= NULL
) {
5523 if ((err
= put_nvlist(zc
, config
)) != 0)
5525 nvlist_free(config
);
5527 nvlist_free(policy
);
5535 * If multihost is enabled, resuming I/O is unsafe as another
5536 * host may have imported the pool.
5538 if (spa_multihost(spa
) && spa_suspended(spa
))
5539 return (SET_ERROR(EINVAL
));
5541 spa_vdev_state_enter(spa
, SCL_NONE
);
5543 if (zc
->zc_guid
== 0) {
5546 vd
= spa_lookup_by_guid(spa
, zc
->zc_guid
, B_TRUE
);
5548 (void) spa_vdev_state_exit(spa
, NULL
, ENODEV
);
5549 spa_close(spa
, FTAG
);
5550 return (SET_ERROR(ENODEV
));
5554 vdev_clear(spa
, vd
);
5556 (void) spa_vdev_state_exit(spa
, spa_suspended(spa
) ?
5557 NULL
: spa
->spa_root_vdev
, 0);
5560 * Resume any suspended I/Os.
5562 if (zio_resume(spa
) != 0)
5563 error
= SET_ERROR(EIO
);
5565 spa_close(spa
, FTAG
);
5571 * Reopen all the vdevs associated with the pool.
5574 * "scrub_restart" -> when true and scrub is running, allow to restart
5575 * scrub as the side effect of the reopen (boolean).
5580 static const zfs_ioc_key_t zfs_keys_pool_reopen
[] = {
5581 {"scrub_restart", DATA_TYPE_BOOLEAN_VALUE
, ZK_OPTIONAL
},
5586 zfs_ioc_pool_reopen(const char *pool
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
5590 boolean_t rc
, scrub_restart
= B_TRUE
;
5593 error
= nvlist_lookup_boolean_value(innvl
,
5594 "scrub_restart", &rc
);
5599 error
= spa_open(pool
, &spa
, FTAG
);
5603 spa_vdev_state_enter(spa
, SCL_NONE
);
5606 * If the scrub_restart flag is B_FALSE and a scrub is already
5607 * in progress then set spa_scrub_reopen flag to B_TRUE so that
5608 * we don't restart the scrub as a side effect of the reopen.
5609 * Otherwise, let vdev_open() decided if a resilver is required.
5612 spa
->spa_scrub_reopen
= (!scrub_restart
&&
5613 dsl_scan_scrubbing(spa
->spa_dsl_pool
));
5614 vdev_reopen(spa
->spa_root_vdev
);
5615 spa
->spa_scrub_reopen
= B_FALSE
;
5617 (void) spa_vdev_state_exit(spa
, NULL
, 0);
5618 spa_close(spa
, FTAG
);
5624 * zc_name name of filesystem
5627 * zc_string name of conflicting snapshot, if there is one
5630 zfs_ioc_promote(zfs_cmd_t
*zc
)
5633 dsl_dataset_t
*ds
, *ods
;
5634 char origin
[ZFS_MAX_DATASET_NAME_LEN
];
5638 zc
->zc_name
[sizeof (zc
->zc_name
) - 1] = '\0';
5639 if (dataset_namecheck(zc
->zc_name
, NULL
, NULL
) != 0 ||
5640 strchr(zc
->zc_name
, '%'))
5641 return (SET_ERROR(EINVAL
));
5643 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
5647 error
= dsl_dataset_hold(dp
, zc
->zc_name
, FTAG
, &ds
);
5649 dsl_pool_rele(dp
, FTAG
);
5653 if (!dsl_dir_is_clone(ds
->ds_dir
)) {
5654 dsl_dataset_rele(ds
, FTAG
);
5655 dsl_pool_rele(dp
, FTAG
);
5656 return (SET_ERROR(EINVAL
));
5659 error
= dsl_dataset_hold_obj(dp
,
5660 dsl_dir_phys(ds
->ds_dir
)->dd_origin_obj
, FTAG
, &ods
);
5662 dsl_dataset_rele(ds
, FTAG
);
5663 dsl_pool_rele(dp
, FTAG
);
5667 dsl_dataset_name(ods
, origin
);
5668 dsl_dataset_rele(ods
, FTAG
);
5669 dsl_dataset_rele(ds
, FTAG
);
5670 dsl_pool_rele(dp
, FTAG
);
5673 * We don't need to unmount *all* the origin fs's snapshots, but
5676 cp
= strchr(origin
, '@');
5679 (void) dmu_objset_find(origin
,
5680 zfs_unmount_snap_cb
, NULL
, DS_FIND_SNAPSHOTS
);
5681 return (dsl_dataset_promote(zc
->zc_name
, zc
->zc_string
));
5685 * Retrieve a single {user|group|project}{used|quota}@... property.
5688 * zc_name name of filesystem
5689 * zc_objset_type zfs_userquota_prop_t
5690 * zc_value domain name (eg. "S-1-234-567-89")
5691 * zc_guid RID/UID/GID
5694 * zc_cookie property value
5697 zfs_ioc_userspace_one(zfs_cmd_t
*zc
)
5702 if (zc
->zc_objset_type
>= ZFS_NUM_USERQUOTA_PROPS
)
5703 return (SET_ERROR(EINVAL
));
5705 error
= zfsvfs_hold(zc
->zc_name
, FTAG
, &zfsvfs
, B_FALSE
);
5709 error
= zfs_userspace_one(zfsvfs
,
5710 zc
->zc_objset_type
, zc
->zc_value
, zc
->zc_guid
, &zc
->zc_cookie
);
5711 zfsvfs_rele(zfsvfs
, FTAG
);
5718 * zc_name name of filesystem
5719 * zc_cookie zap cursor
5720 * zc_objset_type zfs_userquota_prop_t
5721 * zc_nvlist_dst[_size] buffer to fill (not really an nvlist)
5724 * zc_nvlist_dst[_size] data buffer (array of zfs_useracct_t)
5725 * zc_cookie zap cursor
5728 zfs_ioc_userspace_many(zfs_cmd_t
*zc
)
5731 int bufsize
= zc
->zc_nvlist_dst_size
;
5734 return (SET_ERROR(ENOMEM
));
5736 int error
= zfsvfs_hold(zc
->zc_name
, FTAG
, &zfsvfs
, B_FALSE
);
5740 void *buf
= vmem_alloc(bufsize
, KM_SLEEP
);
5742 error
= zfs_userspace_many(zfsvfs
, zc
->zc_objset_type
, &zc
->zc_cookie
,
5743 buf
, &zc
->zc_nvlist_dst_size
);
5746 error
= xcopyout(buf
,
5747 (void *)(uintptr_t)zc
->zc_nvlist_dst
,
5748 zc
->zc_nvlist_dst_size
);
5750 vmem_free(buf
, bufsize
);
5751 zfsvfs_rele(zfsvfs
, FTAG
);
5758 * zc_name name of filesystem
5764 zfs_ioc_userspace_upgrade(zfs_cmd_t
*zc
)
5770 if (getzfsvfs(zc
->zc_name
, &zfsvfs
) == 0) {
5771 if (!dmu_objset_userused_enabled(zfsvfs
->z_os
)) {
5773 * If userused is not enabled, it may be because the
5774 * objset needs to be closed & reopened (to grow the
5775 * objset_phys_t). Suspend/resume the fs will do that.
5777 dsl_dataset_t
*ds
, *newds
;
5779 ds
= dmu_objset_ds(zfsvfs
->z_os
);
5780 error
= zfs_suspend_fs(zfsvfs
);
5782 dmu_objset_refresh_ownership(ds
, &newds
,
5784 error
= zfs_resume_fs(zfsvfs
, newds
);
5788 error
= dmu_objset_userspace_upgrade(zfsvfs
->z_os
);
5789 deactivate_super(zfsvfs
->z_sb
);
5791 /* XXX kind of reading contents without owning */
5792 error
= dmu_objset_hold_flags(zc
->zc_name
, B_TRUE
, FTAG
, &os
);
5796 error
= dmu_objset_userspace_upgrade(os
);
5797 dmu_objset_rele_flags(os
, B_TRUE
, FTAG
);
5805 * zc_name name of filesystem
5811 zfs_ioc_id_quota_upgrade(zfs_cmd_t
*zc
)
5816 error
= dmu_objset_hold_flags(zc
->zc_name
, B_TRUE
, FTAG
, &os
);
5820 if (dmu_objset_userobjspace_upgradable(os
) ||
5821 dmu_objset_projectquota_upgradable(os
)) {
5822 mutex_enter(&os
->os_upgrade_lock
);
5823 if (os
->os_upgrade_id
== 0) {
5824 /* clear potential error code and retry */
5825 os
->os_upgrade_status
= 0;
5826 mutex_exit(&os
->os_upgrade_lock
);
5828 dmu_objset_id_quota_upgrade(os
);
5830 mutex_exit(&os
->os_upgrade_lock
);
5833 dsl_pool_rele(dmu_objset_pool(os
), FTAG
);
5835 taskq_wait_id(os
->os_spa
->spa_upgrade_taskq
, os
->os_upgrade_id
);
5836 error
= os
->os_upgrade_status
;
5838 dsl_pool_rele(dmu_objset_pool(os
), FTAG
);
5841 dsl_dataset_rele_flags(dmu_objset_ds(os
), DS_HOLD_FLAG_DECRYPT
, FTAG
);
5847 zfs_ioc_share(zfs_cmd_t
*zc
)
5849 return (SET_ERROR(ENOSYS
));
5852 ace_t full_access
[] = {
5853 {(uid_t
)-1, ACE_ALL_PERMS
, ACE_EVERYONE
, 0}
5858 * zc_name name of containing filesystem
5859 * zc_obj object # beyond which we want next in-use object #
5862 * zc_obj next in-use object #
5865 zfs_ioc_next_obj(zfs_cmd_t
*zc
)
5867 objset_t
*os
= NULL
;
5870 error
= dmu_objset_hold(zc
->zc_name
, FTAG
, &os
);
5874 error
= dmu_object_next(os
, &zc
->zc_obj
, B_FALSE
, 0);
5876 dmu_objset_rele(os
, FTAG
);
5882 * zc_name name of filesystem
5883 * zc_value prefix name for snapshot
5884 * zc_cleanup_fd cleanup-on-exit file descriptor for calling process
5887 * zc_value short name of new snapshot
5890 zfs_ioc_tmp_snapshot(zfs_cmd_t
*zc
)
5897 error
= zfs_onexit_fd_hold(zc
->zc_cleanup_fd
, &minor
);
5901 snap_name
= kmem_asprintf("%s-%016llx", zc
->zc_value
,
5902 (u_longlong_t
)ddi_get_lbolt64());
5903 hold_name
= kmem_asprintf("%%%s", zc
->zc_value
);
5905 error
= dsl_dataset_snapshot_tmp(zc
->zc_name
, snap_name
, minor
,
5908 (void) strlcpy(zc
->zc_value
, snap_name
,
5909 sizeof (zc
->zc_value
));
5912 zfs_onexit_fd_rele(zc
->zc_cleanup_fd
);
5918 * zc_name name of "to" snapshot
5919 * zc_value name of "from" snapshot
5920 * zc_cookie file descriptor to write diff data on
5923 * dmu_diff_record_t's to the file descriptor
5926 zfs_ioc_diff(zfs_cmd_t
*zc
)
5932 fp
= getf(zc
->zc_cookie
);
5934 return (SET_ERROR(EBADF
));
5938 error
= dmu_diff(zc
->zc_name
, zc
->zc_value
, fp
->f_vnode
, &off
);
5940 if (VOP_SEEK(fp
->f_vnode
, fp
->f_offset
, &off
, NULL
) == 0)
5942 releasef(zc
->zc_cookie
);
5948 zfs_ioc_smb_acl(zfs_cmd_t
*zc
)
5950 return (SET_ERROR(ENOTSUP
));
5955 * "holds" -> { snapname -> holdname (string), ... }
5956 * (optional) "cleanup_fd" -> fd (int32)
5960 * snapname -> error value (int32)
5964 static const zfs_ioc_key_t zfs_keys_hold
[] = {
5965 {"holds", DATA_TYPE_NVLIST
, 0},
5966 {"cleanup_fd", DATA_TYPE_INT32
, ZK_OPTIONAL
},
5971 zfs_ioc_hold(const char *pool
, nvlist_t
*args
, nvlist_t
*errlist
)
5975 int cleanup_fd
= -1;
5979 holds
= fnvlist_lookup_nvlist(args
, "holds");
5981 /* make sure the user didn't pass us any invalid (empty) tags */
5982 for (pair
= nvlist_next_nvpair(holds
, NULL
); pair
!= NULL
;
5983 pair
= nvlist_next_nvpair(holds
, pair
)) {
5986 error
= nvpair_value_string(pair
, &htag
);
5988 return (SET_ERROR(error
));
5990 if (strlen(htag
) == 0)
5991 return (SET_ERROR(EINVAL
));
5994 if (nvlist_lookup_int32(args
, "cleanup_fd", &cleanup_fd
) == 0) {
5995 error
= zfs_onexit_fd_hold(cleanup_fd
, &minor
);
5997 return (SET_ERROR(error
));
6000 error
= dsl_dataset_user_hold(holds
, minor
, errlist
);
6002 zfs_onexit_fd_rele(cleanup_fd
);
6003 return (SET_ERROR(error
));
6007 * innvl is not used.
6010 * holdname -> time added (uint64 seconds since epoch)
6014 static const zfs_ioc_key_t zfs_keys_get_holds
[] = {
6020 zfs_ioc_get_holds(const char *snapname
, nvlist_t
*args
, nvlist_t
*outnvl
)
6022 return (dsl_dataset_get_holds(snapname
, outnvl
));
6027 * snapname -> { holdname, ... }
6032 * snapname -> error value (int32)
6036 static const zfs_ioc_key_t zfs_keys_release
[] = {
6037 {"<snapname>...", DATA_TYPE_NVLIST
, ZK_WILDCARDLIST
},
6042 zfs_ioc_release(const char *pool
, nvlist_t
*holds
, nvlist_t
*errlist
)
6044 return (dsl_dataset_user_release(holds
, errlist
));
6049 * zc_guid flags (ZEVENT_NONBLOCK)
6050 * zc_cleanup_fd zevent file descriptor
6053 * zc_nvlist_dst next nvlist event
6054 * zc_cookie dropped events since last get
6057 zfs_ioc_events_next(zfs_cmd_t
*zc
)
6060 nvlist_t
*event
= NULL
;
6062 uint64_t dropped
= 0;
6065 error
= zfs_zevent_fd_hold(zc
->zc_cleanup_fd
, &minor
, &ze
);
6070 error
= zfs_zevent_next(ze
, &event
,
6071 &zc
->zc_nvlist_dst_size
, &dropped
);
6072 if (event
!= NULL
) {
6073 zc
->zc_cookie
= dropped
;
6074 error
= put_nvlist(zc
, event
);
6078 if (zc
->zc_guid
& ZEVENT_NONBLOCK
)
6081 if ((error
== 0) || (error
!= ENOENT
))
6084 error
= zfs_zevent_wait(ze
);
6089 zfs_zevent_fd_rele(zc
->zc_cleanup_fd
);
6096 * zc_cookie cleared events count
6099 zfs_ioc_events_clear(zfs_cmd_t
*zc
)
6103 zfs_zevent_drain_all(&count
);
6104 zc
->zc_cookie
= count
;
6111 * zc_guid eid | ZEVENT_SEEK_START | ZEVENT_SEEK_END
6112 * zc_cleanup zevent file descriptor
6115 zfs_ioc_events_seek(zfs_cmd_t
*zc
)
6121 error
= zfs_zevent_fd_hold(zc
->zc_cleanup_fd
, &minor
, &ze
);
6125 error
= zfs_zevent_seek(ze
, zc
->zc_guid
);
6126 zfs_zevent_fd_rele(zc
->zc_cleanup_fd
);
6133 * zc_name name of later filesystem or snapshot
6134 * zc_value full name of old snapshot or bookmark
6137 * zc_cookie space in bytes
6138 * zc_objset_type compressed space in bytes
6139 * zc_perm_action uncompressed space in bytes
6142 zfs_ioc_space_written(zfs_cmd_t
*zc
)
6148 error
= dsl_pool_hold(zc
->zc_name
, FTAG
, &dp
);
6151 error
= dsl_dataset_hold(dp
, zc
->zc_name
, FTAG
, &new);
6153 dsl_pool_rele(dp
, FTAG
);
6156 if (strchr(zc
->zc_value
, '#') != NULL
) {
6157 zfs_bookmark_phys_t bmp
;
6158 error
= dsl_bookmark_lookup(dp
, zc
->zc_value
,
6161 error
= dsl_dataset_space_written_bookmark(&bmp
, new,
6163 &zc
->zc_objset_type
, &zc
->zc_perm_action
);
6167 error
= dsl_dataset_hold(dp
, zc
->zc_value
, FTAG
, &old
);
6170 error
= dsl_dataset_space_written(old
, new,
6172 &zc
->zc_objset_type
, &zc
->zc_perm_action
);
6173 dsl_dataset_rele(old
, FTAG
);
6176 dsl_dataset_rele(new, FTAG
);
6177 dsl_pool_rele(dp
, FTAG
);
6183 * "firstsnap" -> snapshot name
6187 * "used" -> space in bytes
6188 * "compressed" -> compressed space in bytes
6189 * "uncompressed" -> uncompressed space in bytes
6192 static const zfs_ioc_key_t zfs_keys_space_snaps
[] = {
6193 {"firstsnap", DATA_TYPE_STRING
, 0},
6197 zfs_ioc_space_snaps(const char *lastsnap
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6201 dsl_dataset_t
*new, *old
;
6203 uint64_t used
, comp
, uncomp
;
6205 firstsnap
= fnvlist_lookup_string(innvl
, "firstsnap");
6207 error
= dsl_pool_hold(lastsnap
, FTAG
, &dp
);
6211 error
= dsl_dataset_hold(dp
, lastsnap
, FTAG
, &new);
6212 if (error
== 0 && !new->ds_is_snapshot
) {
6213 dsl_dataset_rele(new, FTAG
);
6214 error
= SET_ERROR(EINVAL
);
6217 dsl_pool_rele(dp
, FTAG
);
6220 error
= dsl_dataset_hold(dp
, firstsnap
, FTAG
, &old
);
6221 if (error
== 0 && !old
->ds_is_snapshot
) {
6222 dsl_dataset_rele(old
, FTAG
);
6223 error
= SET_ERROR(EINVAL
);
6226 dsl_dataset_rele(new, FTAG
);
6227 dsl_pool_rele(dp
, FTAG
);
6231 error
= dsl_dataset_space_wouldfree(old
, new, &used
, &comp
, &uncomp
);
6232 dsl_dataset_rele(old
, FTAG
);
6233 dsl_dataset_rele(new, FTAG
);
6234 dsl_pool_rele(dp
, FTAG
);
6235 fnvlist_add_uint64(outnvl
, "used", used
);
6236 fnvlist_add_uint64(outnvl
, "compressed", comp
);
6237 fnvlist_add_uint64(outnvl
, "uncompressed", uncomp
);
6243 * "fd" -> file descriptor to write stream to (int32)
6244 * (optional) "fromsnap" -> full snap name to send an incremental from
6245 * (optional) "largeblockok" -> (value ignored)
6246 * indicates that blocks > 128KB are permitted
6247 * (optional) "embedok" -> (value ignored)
6248 * presence indicates DRR_WRITE_EMBEDDED records are permitted
6249 * (optional) "compressok" -> (value ignored)
6250 * presence indicates compressed DRR_WRITE records are permitted
6251 * (optional) "rawok" -> (value ignored)
6252 * presence indicates raw encrypted records should be used.
6253 * (optional) "resume_object" and "resume_offset" -> (uint64)
6254 * if present, resume send stream from specified object and offset.
6255 * (optional) "redactbook" -> (string)
6256 * if present, use this bookmark's redaction list to generate a redacted
6262 static const zfs_ioc_key_t zfs_keys_send_new
[] = {
6263 {"fd", DATA_TYPE_INT32
, 0},
6264 {"fromsnap", DATA_TYPE_STRING
, ZK_OPTIONAL
},
6265 {"largeblockok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6266 {"embedok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6267 {"compressok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6268 {"rawok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6269 {"resume_object", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6270 {"resume_offset", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6271 {"redactbook", DATA_TYPE_STRING
, ZK_OPTIONAL
},
6276 zfs_ioc_send_new(const char *snapname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6280 char *fromname
= NULL
;
6283 boolean_t largeblockok
;
6285 boolean_t compressok
;
6287 uint64_t resumeobj
= 0;
6288 uint64_t resumeoff
= 0;
6289 char *redactbook
= NULL
;
6291 fd
= fnvlist_lookup_int32(innvl
, "fd");
6293 (void) nvlist_lookup_string(innvl
, "fromsnap", &fromname
);
6295 largeblockok
= nvlist_exists(innvl
, "largeblockok");
6296 embedok
= nvlist_exists(innvl
, "embedok");
6297 compressok
= nvlist_exists(innvl
, "compressok");
6298 rawok
= nvlist_exists(innvl
, "rawok");
6300 (void) nvlist_lookup_uint64(innvl
, "resume_object", &resumeobj
);
6301 (void) nvlist_lookup_uint64(innvl
, "resume_offset", &resumeoff
);
6303 (void) nvlist_lookup_string(innvl
, "redactbook", &redactbook
);
6305 if ((fp
= getf(fd
)) == NULL
)
6306 return (SET_ERROR(EBADF
));
6309 dmu_send_outparams_t out
= {0};
6310 out
.dso_outfunc
= dump_bytes
;
6311 out
.dso_arg
= fp
->f_vnode
;
6312 out
.dso_dryrun
= B_FALSE
;
6313 error
= dmu_send(snapname
, fromname
, embedok
, largeblockok
, compressok
,
6314 rawok
, resumeobj
, resumeoff
, redactbook
, fd
, &off
, &out
);
6316 if (VOP_SEEK(fp
->f_vnode
, fp
->f_offset
, &off
, NULL
) == 0)
6325 send_space_sum(objset_t
*os
, void *buf
, int len
, void *arg
)
6327 uint64_t *size
= arg
;
6333 * Determine approximately how large a zfs send stream will be -- the number
6334 * of bytes that will be written to the fd supplied to zfs_ioc_send_new().
6337 * (optional) "from" -> full snap or bookmark name to send an incremental
6339 * (optional) "largeblockok" -> (value ignored)
6340 * indicates that blocks > 128KB are permitted
6341 * (optional) "embedok" -> (value ignored)
6342 * presence indicates DRR_WRITE_EMBEDDED records are permitted
6343 * (optional) "compressok" -> (value ignored)
6344 * presence indicates compressed DRR_WRITE records are permitted
6345 * (optional) "rawok" -> (value ignored)
6346 * presence indicates raw encrypted records should be used.
6347 * (optional) "fd" -> file descriptor to use as a cookie for progress
6352 * "space" -> bytes of space (uint64)
6355 static const zfs_ioc_key_t zfs_keys_send_space
[] = {
6356 {"from", DATA_TYPE_STRING
, ZK_OPTIONAL
},
6357 {"fromsnap", DATA_TYPE_STRING
, ZK_OPTIONAL
},
6358 {"largeblockok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6359 {"embedok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6360 {"compressok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6361 {"rawok", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6362 {"fd", DATA_TYPE_INT32
, ZK_OPTIONAL
},
6363 {"redactbook", DATA_TYPE_STRING
, ZK_OPTIONAL
},
6364 {"resumeobj", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6365 {"resumeoff", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6366 {"bytes", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6370 zfs_ioc_send_space(const char *snapname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6373 dsl_dataset_t
*tosnap
;
6374 dsl_dataset_t
*fromsnap
= NULL
;
6376 char *fromname
= NULL
;
6377 char *redactlist_book
= NULL
;
6378 boolean_t largeblockok
;
6380 boolean_t compressok
;
6383 boolean_t full_estimate
= B_FALSE
;
6384 uint64_t resumeobj
= 0;
6385 uint64_t resumeoff
= 0;
6386 uint64_t resume_bytes
= 0;
6388 zfs_bookmark_phys_t zbm
= {0};
6390 error
= dsl_pool_hold(snapname
, FTAG
, &dp
);
6394 error
= dsl_dataset_hold(dp
, snapname
, FTAG
, &tosnap
);
6396 dsl_pool_rele(dp
, FTAG
);
6399 (void) nvlist_lookup_int32(innvl
, "fd", &fd
);
6401 largeblockok
= nvlist_exists(innvl
, "largeblockok");
6402 embedok
= nvlist_exists(innvl
, "embedok");
6403 compressok
= nvlist_exists(innvl
, "compressok");
6404 rawok
= nvlist_exists(innvl
, "rawok");
6405 boolean_t from
= (nvlist_lookup_string(innvl
, "from", &fromname
) == 0);
6406 boolean_t altbook
= (nvlist_lookup_string(innvl
, "redactbook",
6407 &redactlist_book
) == 0);
6409 (void) nvlist_lookup_uint64(innvl
, "resume_object", &resumeobj
);
6410 (void) nvlist_lookup_uint64(innvl
, "resume_offset", &resumeoff
);
6411 (void) nvlist_lookup_uint64(innvl
, "bytes", &resume_bytes
);
6414 full_estimate
= B_TRUE
;
6416 if (strchr(fromname
, '#')) {
6417 error
= dsl_bookmark_lookup(dp
, fromname
, tosnap
, &zbm
);
6420 * dsl_bookmark_lookup() will fail with EXDEV if
6421 * the from-bookmark and tosnap are at the same txg.
6422 * However, it's valid to do a send (and therefore,
6423 * a send estimate) from and to the same time point,
6424 * if the bookmark is redacted (the incremental send
6425 * can change what's redacted on the target). In
6426 * this case, dsl_bookmark_lookup() fills in zbm
6427 * but returns EXDEV. Ignore this error.
6429 if (error
== EXDEV
&& zbm
.zbm_redaction_obj
!= 0 &&
6431 dsl_dataset_phys(tosnap
)->ds_guid
)
6435 dsl_dataset_rele(tosnap
, FTAG
);
6436 dsl_pool_rele(dp
, FTAG
);
6439 if (zbm
.zbm_redaction_obj
!= 0 || !(zbm
.zbm_flags
&
6440 ZBM_FLAG_HAS_FBN
)) {
6441 full_estimate
= B_TRUE
;
6443 } else if (strchr(fromname
, '@')) {
6444 error
= dsl_dataset_hold(dp
, fromname
, FTAG
, &fromsnap
);
6446 dsl_dataset_rele(tosnap
, FTAG
);
6447 dsl_pool_rele(dp
, FTAG
);
6451 if (!dsl_dataset_is_before(tosnap
, fromsnap
, 0)) {
6452 full_estimate
= B_TRUE
;
6453 dsl_dataset_rele(fromsnap
, FTAG
);
6457 * from is not properly formatted as a snapshot or
6460 dsl_dataset_rele(tosnap
, FTAG
);
6461 dsl_pool_rele(dp
, FTAG
);
6462 return (SET_ERROR(EINVAL
));
6466 if (full_estimate
) {
6467 dmu_send_outparams_t out
= {0};
6469 out
.dso_outfunc
= send_space_sum
;
6470 out
.dso_arg
= &space
;
6471 out
.dso_dryrun
= B_TRUE
;
6473 * We have to release these holds so dmu_send can take them. It
6474 * will do all the error checking we need.
6476 dsl_dataset_rele(tosnap
, FTAG
);
6477 dsl_pool_rele(dp
, FTAG
);
6478 error
= dmu_send(snapname
, fromname
, embedok
, largeblockok
,
6479 compressok
, rawok
, resumeobj
, resumeoff
, redactlist_book
,
6482 error
= dmu_send_estimate_fast(tosnap
, fromsnap
,
6483 (from
&& strchr(fromname
, '#') != NULL
? &zbm
: NULL
),
6484 compressok
|| rawok
, &space
);
6485 space
-= resume_bytes
;
6486 if (fromsnap
!= NULL
)
6487 dsl_dataset_rele(fromsnap
, FTAG
);
6488 dsl_dataset_rele(tosnap
, FTAG
);
6489 dsl_pool_rele(dp
, FTAG
);
6492 fnvlist_add_uint64(outnvl
, "space", space
);
6498 * Sync the currently open TXG to disk for the specified pool.
6499 * This is somewhat similar to 'zfs_sync()'.
6500 * For cases that do not result in error this ioctl will wait for
6501 * the currently open TXG to commit before returning back to the caller.
6504 * "force" -> when true, force uberblock update even if there is no dirty data.
6505 * In addition this will cause the vdev configuration to be written
6506 * out including updating the zpool cache file. (boolean_t)
6511 static const zfs_ioc_key_t zfs_keys_pool_sync
[] = {
6512 {"force", DATA_TYPE_BOOLEAN_VALUE
, 0},
6517 zfs_ioc_pool_sync(const char *pool
, nvlist_t
*innvl
, nvlist_t
*onvl
)
6520 boolean_t force
= B_FALSE
;
6523 if ((err
= spa_open(pool
, &spa
, FTAG
)) != 0)
6527 force
= fnvlist_lookup_boolean_value(innvl
, "force");
6530 spa_config_enter(spa
, SCL_CONFIG
, FTAG
, RW_WRITER
);
6531 vdev_config_dirty(spa
->spa_root_vdev
);
6532 spa_config_exit(spa
, SCL_CONFIG
, FTAG
);
6534 txg_wait_synced(spa_get_dsl(spa
), 0);
6536 spa_close(spa
, FTAG
);
6542 * Load a user's wrapping key into the kernel.
6544 * "hidden_args" -> { "wkeydata" -> value }
6545 * raw uint8_t array of encryption wrapping key data (32 bytes)
6546 * (optional) "noop" -> (value ignored)
6547 * presence indicated key should only be verified, not loaded
6550 static const zfs_ioc_key_t zfs_keys_load_key
[] = {
6551 {"hidden_args", DATA_TYPE_NVLIST
, 0},
6552 {"noop", DATA_TYPE_BOOLEAN
, ZK_OPTIONAL
},
6557 zfs_ioc_load_key(const char *dsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6560 dsl_crypto_params_t
*dcp
= NULL
;
6561 nvlist_t
*hidden_args
;
6562 boolean_t noop
= nvlist_exists(innvl
, "noop");
6564 if (strchr(dsname
, '@') != NULL
|| strchr(dsname
, '%') != NULL
) {
6565 ret
= SET_ERROR(EINVAL
);
6569 hidden_args
= fnvlist_lookup_nvlist(innvl
, ZPOOL_HIDDEN_ARGS
);
6571 ret
= dsl_crypto_params_create_nvlist(DCP_CMD_NONE
, NULL
,
6576 ret
= spa_keystore_load_wkey(dsname
, dcp
, noop
);
6580 dsl_crypto_params_free(dcp
, noop
);
6585 dsl_crypto_params_free(dcp
, B_TRUE
);
6590 * Unload a user's wrapping key from the kernel.
6591 * Both innvl and outnvl are unused.
6593 static const zfs_ioc_key_t zfs_keys_unload_key
[] = {
6599 zfs_ioc_unload_key(const char *dsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6603 if (strchr(dsname
, '@') != NULL
|| strchr(dsname
, '%') != NULL
) {
6604 ret
= (SET_ERROR(EINVAL
));
6608 ret
= spa_keystore_unload_wkey(dsname
);
6617 * Changes a user's wrapping key used to decrypt a dataset. The keyformat,
6618 * keylocation, pbkdf2salt, and pbkdf2iters properties can also be specified
6619 * here to change how the key is derived in userspace.
6622 * "hidden_args" (optional) -> { "wkeydata" -> value }
6623 * raw uint8_t array of new encryption wrapping key data (32 bytes)
6624 * "props" (optional) -> { prop -> value }
6629 static const zfs_ioc_key_t zfs_keys_change_key
[] = {
6630 {"crypt_cmd", DATA_TYPE_UINT64
, ZK_OPTIONAL
},
6631 {"hidden_args", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
6632 {"props", DATA_TYPE_NVLIST
, ZK_OPTIONAL
},
6637 zfs_ioc_change_key(const char *dsname
, nvlist_t
*innvl
, nvlist_t
*outnvl
)
6640 uint64_t cmd
= DCP_CMD_NONE
;
6641 dsl_crypto_params_t
*dcp
= NULL
;
6642 nvlist_t
*args
= NULL
, *hidden_args
= NULL
;
6644 if (strchr(dsname
, '@') != NULL
|| strchr(dsname
, '%') != NULL
) {
6645 ret
= (SET_ERROR(EINVAL
));
6649 (void) nvlist_lookup_uint64(innvl
, "crypt_cmd", &cmd
);
6650 (void) nvlist_lookup_nvlist(innvl
, "props", &args
);
6651 (void) nvlist_lookup_nvlist(innvl
, ZPOOL_HIDDEN_ARGS
, &hidden_args
);
6653 ret
= dsl_crypto_params_create_nvlist(cmd
, args
, hidden_args
, &dcp
);
6657 ret
= spa_keystore_change_key(dsname
, dcp
);
6661 dsl_crypto_params_free(dcp
, B_FALSE
);
6666 dsl_crypto_params_free(dcp
, B_TRUE
);
6670 static zfs_ioc_vec_t zfs_ioc_vec
[ZFS_IOC_LAST
- ZFS_IOC_FIRST
];
6673 zfs_ioctl_register_legacy(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
,
6674 zfs_secpolicy_func_t
*secpolicy
, zfs_ioc_namecheck_t namecheck
,
6675 boolean_t log_history
, zfs_ioc_poolcheck_t pool_check
)
6677 zfs_ioc_vec_t
*vec
= &zfs_ioc_vec
[ioc
- ZFS_IOC_FIRST
];
6679 ASSERT3U(ioc
, >=, ZFS_IOC_FIRST
);
6680 ASSERT3U(ioc
, <, ZFS_IOC_LAST
);
6681 ASSERT3P(vec
->zvec_legacy_func
, ==, NULL
);
6682 ASSERT3P(vec
->zvec_func
, ==, NULL
);
6684 vec
->zvec_legacy_func
= func
;
6685 vec
->zvec_secpolicy
= secpolicy
;
6686 vec
->zvec_namecheck
= namecheck
;
6687 vec
->zvec_allow_log
= log_history
;
6688 vec
->zvec_pool_check
= pool_check
;
6692 * See the block comment at the beginning of this file for details on
6693 * each argument to this function.
6696 zfs_ioctl_register(const char *name
, zfs_ioc_t ioc
, zfs_ioc_func_t
*func
,
6697 zfs_secpolicy_func_t
*secpolicy
, zfs_ioc_namecheck_t namecheck
,
6698 zfs_ioc_poolcheck_t pool_check
, boolean_t smush_outnvlist
,
6699 boolean_t allow_log
, const zfs_ioc_key_t
*nvl_keys
, size_t num_keys
)
6701 zfs_ioc_vec_t
*vec
= &zfs_ioc_vec
[ioc
- ZFS_IOC_FIRST
];
6703 ASSERT3U(ioc
, >=, ZFS_IOC_FIRST
);
6704 ASSERT3U(ioc
, <, ZFS_IOC_LAST
);
6705 ASSERT3P(vec
->zvec_legacy_func
, ==, NULL
);
6706 ASSERT3P(vec
->zvec_func
, ==, NULL
);
6708 /* if we are logging, the name must be valid */
6709 ASSERT(!allow_log
|| namecheck
!= NO_NAME
);
6711 vec
->zvec_name
= name
;
6712 vec
->zvec_func
= func
;
6713 vec
->zvec_secpolicy
= secpolicy
;
6714 vec
->zvec_namecheck
= namecheck
;
6715 vec
->zvec_pool_check
= pool_check
;
6716 vec
->zvec_smush_outnvlist
= smush_outnvlist
;
6717 vec
->zvec_allow_log
= allow_log
;
6718 vec
->zvec_nvl_keys
= nvl_keys
;
6719 vec
->zvec_nvl_key_count
= num_keys
;
6723 zfs_ioctl_register_pool(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
,
6724 zfs_secpolicy_func_t
*secpolicy
, boolean_t log_history
,
6725 zfs_ioc_poolcheck_t pool_check
)
6727 zfs_ioctl_register_legacy(ioc
, func
, secpolicy
,
6728 POOL_NAME
, log_history
, pool_check
);
6732 zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
,
6733 zfs_secpolicy_func_t
*secpolicy
, zfs_ioc_poolcheck_t pool_check
)
6735 zfs_ioctl_register_legacy(ioc
, func
, secpolicy
,
6736 DATASET_NAME
, B_FALSE
, pool_check
);
6740 zfs_ioctl_register_pool_modify(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
)
6742 zfs_ioctl_register_legacy(ioc
, func
, zfs_secpolicy_config
,
6743 POOL_NAME
, B_TRUE
, POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
);
6747 zfs_ioctl_register_pool_meta(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
,
6748 zfs_secpolicy_func_t
*secpolicy
)
6750 zfs_ioctl_register_legacy(ioc
, func
, secpolicy
,
6751 NO_NAME
, B_FALSE
, POOL_CHECK_NONE
);
6755 zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc
,
6756 zfs_ioc_legacy_func_t
*func
, zfs_secpolicy_func_t
*secpolicy
)
6758 zfs_ioctl_register_legacy(ioc
, func
, secpolicy
,
6759 DATASET_NAME
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6763 zfs_ioctl_register_dataset_read(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
)
6765 zfs_ioctl_register_dataset_read_secpolicy(ioc
, func
,
6766 zfs_secpolicy_read
);
6770 zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc
, zfs_ioc_legacy_func_t
*func
,
6771 zfs_secpolicy_func_t
*secpolicy
)
6773 zfs_ioctl_register_legacy(ioc
, func
, secpolicy
,
6774 DATASET_NAME
, B_TRUE
, POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
);
6778 zfs_ioctl_init(void)
6780 zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT
,
6781 zfs_ioc_snapshot
, zfs_secpolicy_snapshot
, POOL_NAME
,
6782 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6783 zfs_keys_snapshot
, ARRAY_SIZE(zfs_keys_snapshot
));
6785 zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY
,
6786 zfs_ioc_log_history
, zfs_secpolicy_log_history
, NO_NAME
,
6787 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_FALSE
, B_FALSE
,
6788 zfs_keys_log_history
, ARRAY_SIZE(zfs_keys_log_history
));
6790 zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS
,
6791 zfs_ioc_space_snaps
, zfs_secpolicy_read
, DATASET_NAME
,
6792 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
,
6793 zfs_keys_space_snaps
, ARRAY_SIZE(zfs_keys_space_snaps
));
6795 zfs_ioctl_register("send", ZFS_IOC_SEND_NEW
,
6796 zfs_ioc_send_new
, zfs_secpolicy_send_new
, DATASET_NAME
,
6797 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
,
6798 zfs_keys_send_new
, ARRAY_SIZE(zfs_keys_send_new
));
6800 zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE
,
6801 zfs_ioc_send_space
, zfs_secpolicy_read
, DATASET_NAME
,
6802 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
,
6803 zfs_keys_send_space
, ARRAY_SIZE(zfs_keys_send_space
));
6805 zfs_ioctl_register("create", ZFS_IOC_CREATE
,
6806 zfs_ioc_create
, zfs_secpolicy_create_clone
, DATASET_NAME
,
6807 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6808 zfs_keys_create
, ARRAY_SIZE(zfs_keys_create
));
6810 zfs_ioctl_register("clone", ZFS_IOC_CLONE
,
6811 zfs_ioc_clone
, zfs_secpolicy_create_clone
, DATASET_NAME
,
6812 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6813 zfs_keys_clone
, ARRAY_SIZE(zfs_keys_clone
));
6815 zfs_ioctl_register("remap", ZFS_IOC_REMAP
,
6816 zfs_ioc_remap
, zfs_secpolicy_none
, DATASET_NAME
,
6817 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_FALSE
, B_TRUE
,
6818 zfs_keys_remap
, ARRAY_SIZE(zfs_keys_remap
));
6820 zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS
,
6821 zfs_ioc_destroy_snaps
, zfs_secpolicy_destroy_snaps
, POOL_NAME
,
6822 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6823 zfs_keys_destroy_snaps
, ARRAY_SIZE(zfs_keys_destroy_snaps
));
6825 zfs_ioctl_register("hold", ZFS_IOC_HOLD
,
6826 zfs_ioc_hold
, zfs_secpolicy_hold
, POOL_NAME
,
6827 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6828 zfs_keys_hold
, ARRAY_SIZE(zfs_keys_hold
));
6829 zfs_ioctl_register("release", ZFS_IOC_RELEASE
,
6830 zfs_ioc_release
, zfs_secpolicy_release
, POOL_NAME
,
6831 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6832 zfs_keys_release
, ARRAY_SIZE(zfs_keys_release
));
6834 zfs_ioctl_register("get_holds", ZFS_IOC_GET_HOLDS
,
6835 zfs_ioc_get_holds
, zfs_secpolicy_read
, DATASET_NAME
,
6836 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
,
6837 zfs_keys_get_holds
, ARRAY_SIZE(zfs_keys_get_holds
));
6839 zfs_ioctl_register("rollback", ZFS_IOC_ROLLBACK
,
6840 zfs_ioc_rollback
, zfs_secpolicy_rollback
, DATASET_NAME
,
6841 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_FALSE
, B_TRUE
,
6842 zfs_keys_rollback
, ARRAY_SIZE(zfs_keys_rollback
));
6844 zfs_ioctl_register("bookmark", ZFS_IOC_BOOKMARK
,
6845 zfs_ioc_bookmark
, zfs_secpolicy_bookmark
, POOL_NAME
,
6846 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6847 zfs_keys_bookmark
, ARRAY_SIZE(zfs_keys_bookmark
));
6849 zfs_ioctl_register("get_bookmarks", ZFS_IOC_GET_BOOKMARKS
,
6850 zfs_ioc_get_bookmarks
, zfs_secpolicy_read
, DATASET_NAME
,
6851 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
,
6852 zfs_keys_get_bookmarks
, ARRAY_SIZE(zfs_keys_get_bookmarks
));
6854 zfs_ioctl_register("get_bookmark_props", ZFS_IOC_GET_BOOKMARK_PROPS
,
6855 zfs_ioc_get_bookmark_props
, zfs_secpolicy_read
, ENTITY_NAME
,
6856 POOL_CHECK_SUSPENDED
, B_FALSE
, B_FALSE
, zfs_keys_get_bookmark_props
,
6857 ARRAY_SIZE(zfs_keys_get_bookmark_props
));
6859 zfs_ioctl_register("destroy_bookmarks", ZFS_IOC_DESTROY_BOOKMARKS
,
6860 zfs_ioc_destroy_bookmarks
, zfs_secpolicy_destroy_bookmarks
,
6862 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6863 zfs_keys_destroy_bookmarks
,
6864 ARRAY_SIZE(zfs_keys_destroy_bookmarks
));
6866 zfs_ioctl_register("receive", ZFS_IOC_RECV_NEW
,
6867 zfs_ioc_recv_new
, zfs_secpolicy_recv_new
, DATASET_NAME
,
6868 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6869 zfs_keys_recv_new
, ARRAY_SIZE(zfs_keys_recv_new
));
6870 zfs_ioctl_register("load-key", ZFS_IOC_LOAD_KEY
,
6871 zfs_ioc_load_key
, zfs_secpolicy_load_key
,
6872 DATASET_NAME
, POOL_CHECK_SUSPENDED
, B_TRUE
, B_TRUE
,
6873 zfs_keys_load_key
, ARRAY_SIZE(zfs_keys_load_key
));
6874 zfs_ioctl_register("unload-key", ZFS_IOC_UNLOAD_KEY
,
6875 zfs_ioc_unload_key
, zfs_secpolicy_load_key
,
6876 DATASET_NAME
, POOL_CHECK_SUSPENDED
, B_TRUE
, B_TRUE
,
6877 zfs_keys_unload_key
, ARRAY_SIZE(zfs_keys_unload_key
));
6878 zfs_ioctl_register("change-key", ZFS_IOC_CHANGE_KEY
,
6879 zfs_ioc_change_key
, zfs_secpolicy_change_key
,
6880 DATASET_NAME
, POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
,
6881 B_TRUE
, B_TRUE
, zfs_keys_change_key
,
6882 ARRAY_SIZE(zfs_keys_change_key
));
6884 zfs_ioctl_register("sync", ZFS_IOC_POOL_SYNC
,
6885 zfs_ioc_pool_sync
, zfs_secpolicy_none
, POOL_NAME
,
6886 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_FALSE
, B_FALSE
,
6887 zfs_keys_pool_sync
, ARRAY_SIZE(zfs_keys_pool_sync
));
6888 zfs_ioctl_register("reopen", ZFS_IOC_POOL_REOPEN
, zfs_ioc_pool_reopen
,
6889 zfs_secpolicy_config
, POOL_NAME
, POOL_CHECK_SUSPENDED
, B_TRUE
,
6890 B_TRUE
, zfs_keys_pool_reopen
, ARRAY_SIZE(zfs_keys_pool_reopen
));
6892 zfs_ioctl_register("channel_program", ZFS_IOC_CHANNEL_PROGRAM
,
6893 zfs_ioc_channel_program
, zfs_secpolicy_config
,
6894 POOL_NAME
, POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
,
6895 B_TRUE
, zfs_keys_channel_program
,
6896 ARRAY_SIZE(zfs_keys_channel_program
));
6898 zfs_ioctl_register("redact", ZFS_IOC_REDACT
,
6899 zfs_ioc_redact
, zfs_secpolicy_config
, DATASET_NAME
,
6900 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6901 zfs_keys_redact
, ARRAY_SIZE(zfs_keys_redact
));
6903 zfs_ioctl_register("zpool_checkpoint", ZFS_IOC_POOL_CHECKPOINT
,
6904 zfs_ioc_pool_checkpoint
, zfs_secpolicy_config
, POOL_NAME
,
6905 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6906 zfs_keys_pool_checkpoint
, ARRAY_SIZE(zfs_keys_pool_checkpoint
));
6908 zfs_ioctl_register("zpool_discard_checkpoint",
6909 ZFS_IOC_POOL_DISCARD_CHECKPOINT
, zfs_ioc_pool_discard_checkpoint
,
6910 zfs_secpolicy_config
, POOL_NAME
,
6911 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6912 zfs_keys_pool_discard_checkpoint
,
6913 ARRAY_SIZE(zfs_keys_pool_discard_checkpoint
));
6915 zfs_ioctl_register("initialize", ZFS_IOC_POOL_INITIALIZE
,
6916 zfs_ioc_pool_initialize
, zfs_secpolicy_config
, POOL_NAME
,
6917 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6918 zfs_keys_pool_initialize
, ARRAY_SIZE(zfs_keys_pool_initialize
));
6920 zfs_ioctl_register("trim", ZFS_IOC_POOL_TRIM
,
6921 zfs_ioc_pool_trim
, zfs_secpolicy_config
, POOL_NAME
,
6922 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_TRUE
, B_TRUE
,
6923 zfs_keys_pool_trim
, ARRAY_SIZE(zfs_keys_pool_trim
));
6925 zfs_ioctl_register("wait", ZFS_IOC_WAIT
,
6926 zfs_ioc_wait
, zfs_secpolicy_none
, POOL_NAME
,
6927 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
, B_FALSE
, B_FALSE
,
6928 zfs_keys_pool_wait
, ARRAY_SIZE(zfs_keys_pool_wait
));
6930 /* IOCTLS that use the legacy function signature */
6932 zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE
, zfs_ioc_pool_freeze
,
6933 zfs_secpolicy_config
, NO_NAME
, B_FALSE
, POOL_CHECK_READONLY
);
6935 zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE
, zfs_ioc_pool_create
,
6936 zfs_secpolicy_config
, B_TRUE
, POOL_CHECK_NONE
);
6937 zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN
,
6939 zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE
,
6940 zfs_ioc_pool_upgrade
);
6941 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD
,
6943 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE
,
6944 zfs_ioc_vdev_remove
);
6945 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE
,
6946 zfs_ioc_vdev_set_state
);
6947 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH
,
6948 zfs_ioc_vdev_attach
);
6949 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH
,
6950 zfs_ioc_vdev_detach
);
6951 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH
,
6952 zfs_ioc_vdev_setpath
);
6953 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU
,
6954 zfs_ioc_vdev_setfru
);
6955 zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS
,
6956 zfs_ioc_pool_set_props
);
6957 zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT
,
6958 zfs_ioc_vdev_split
);
6959 zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID
,
6960 zfs_ioc_pool_reguid
);
6962 zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS
,
6963 zfs_ioc_pool_configs
, zfs_secpolicy_none
);
6964 zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT
,
6965 zfs_ioc_pool_tryimport
, zfs_secpolicy_config
);
6966 zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT
,
6967 zfs_ioc_inject_fault
, zfs_secpolicy_inject
);
6968 zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT
,
6969 zfs_ioc_clear_fault
, zfs_secpolicy_inject
);
6970 zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT
,
6971 zfs_ioc_inject_list_next
, zfs_secpolicy_inject
);
6974 * pool destroy, and export don't log the history as part of
6975 * zfsdev_ioctl, but rather zfs_ioc_pool_export
6976 * does the logging of those commands.
6978 zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY
, zfs_ioc_pool_destroy
,
6979 zfs_secpolicy_config
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6980 zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT
, zfs_ioc_pool_export
,
6981 zfs_secpolicy_config
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6983 zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS
, zfs_ioc_pool_stats
,
6984 zfs_secpolicy_read
, B_FALSE
, POOL_CHECK_NONE
);
6985 zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS
, zfs_ioc_pool_get_props
,
6986 zfs_secpolicy_read
, B_FALSE
, POOL_CHECK_NONE
);
6988 zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG
, zfs_ioc_error_log
,
6989 zfs_secpolicy_inject
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6990 zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME
,
6991 zfs_ioc_dsobj_to_dsname
,
6992 zfs_secpolicy_diff
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6993 zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY
,
6994 zfs_ioc_pool_get_history
,
6995 zfs_secpolicy_config
, B_FALSE
, POOL_CHECK_SUSPENDED
);
6997 zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT
, zfs_ioc_pool_import
,
6998 zfs_secpolicy_config
, B_TRUE
, POOL_CHECK_NONE
);
7000 zfs_ioctl_register_pool(ZFS_IOC_CLEAR
, zfs_ioc_clear
,
7001 zfs_secpolicy_config
, B_TRUE
, POOL_CHECK_READONLY
);
7003 zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN
,
7004 zfs_ioc_space_written
);
7005 zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS
,
7006 zfs_ioc_objset_recvd_props
);
7007 zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ
,
7009 zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL
,
7011 zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS
,
7012 zfs_ioc_objset_stats
);
7013 zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS
,
7014 zfs_ioc_objset_zplprops
);
7015 zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT
,
7016 zfs_ioc_dataset_list_next
);
7017 zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT
,
7018 zfs_ioc_snapshot_list_next
);
7019 zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS
,
7020 zfs_ioc_send_progress
);
7022 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF
,
7023 zfs_ioc_diff
, zfs_secpolicy_diff
);
7024 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS
,
7025 zfs_ioc_obj_to_stats
, zfs_secpolicy_diff
);
7026 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH
,
7027 zfs_ioc_obj_to_path
, zfs_secpolicy_diff
);
7028 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE
,
7029 zfs_ioc_userspace_one
, zfs_secpolicy_userspace_one
);
7030 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY
,
7031 zfs_ioc_userspace_many
, zfs_secpolicy_userspace_many
);
7032 zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND
,
7033 zfs_ioc_send
, zfs_secpolicy_send
);
7035 zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP
, zfs_ioc_set_prop
,
7036 zfs_secpolicy_none
);
7037 zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY
, zfs_ioc_destroy
,
7038 zfs_secpolicy_destroy
);
7039 zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME
, zfs_ioc_rename
,
7040 zfs_secpolicy_rename
);
7041 zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV
, zfs_ioc_recv
,
7042 zfs_secpolicy_recv
);
7043 zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE
, zfs_ioc_promote
,
7044 zfs_secpolicy_promote
);
7045 zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP
,
7046 zfs_ioc_inherit_prop
, zfs_secpolicy_inherit_prop
);
7047 zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL
, zfs_ioc_set_fsacl
,
7048 zfs_secpolicy_set_fsacl
);
7050 zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE
, zfs_ioc_share
,
7051 zfs_secpolicy_share
, POOL_CHECK_NONE
);
7052 zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL
, zfs_ioc_smb_acl
,
7053 zfs_secpolicy_smb_acl
, POOL_CHECK_NONE
);
7054 zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE
,
7055 zfs_ioc_userspace_upgrade
, zfs_secpolicy_userspace_upgrade
,
7056 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
);
7057 zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT
,
7058 zfs_ioc_tmp_snapshot
, zfs_secpolicy_tmp_snapshot
,
7059 POOL_CHECK_SUSPENDED
| POOL_CHECK_READONLY
);
7061 zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_NEXT
, zfs_ioc_events_next
,
7062 zfs_secpolicy_config
, NO_NAME
, B_FALSE
, POOL_CHECK_NONE
);
7063 zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_CLEAR
, zfs_ioc_events_clear
,
7064 zfs_secpolicy_config
, NO_NAME
, B_FALSE
, POOL_CHECK_NONE
);
7065 zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_SEEK
, zfs_ioc_events_seek
,
7066 zfs_secpolicy_config
, NO_NAME
, B_FALSE
, POOL_CHECK_NONE
);
7068 zfs_ioctl_init_os();
7072 * Verify that for non-legacy ioctls the input nvlist
7073 * pairs match against the expected input.
7075 * Possible errors are:
7076 * ZFS_ERR_IOC_ARG_UNAVAIL An unrecognized nvpair was encountered
7077 * ZFS_ERR_IOC_ARG_REQUIRED A required nvpair is missing
7078 * ZFS_ERR_IOC_ARG_BADTYPE Invalid type for nvpair
7081 zfs_check_input_nvpairs(nvlist_t
*innvl
, const zfs_ioc_vec_t
*vec
)
7083 const zfs_ioc_key_t
*nvl_keys
= vec
->zvec_nvl_keys
;
7084 boolean_t required_keys_found
= B_FALSE
;
7087 * examine each input pair
7089 for (nvpair_t
*pair
= nvlist_next_nvpair(innvl
, NULL
);
7090 pair
!= NULL
; pair
= nvlist_next_nvpair(innvl
, pair
)) {
7091 char *name
= nvpair_name(pair
);
7092 data_type_t type
= nvpair_type(pair
);
7093 boolean_t identified
= B_FALSE
;
7096 * check pair against the documented names and type
7098 for (int k
= 0; k
< vec
->zvec_nvl_key_count
; k
++) {
7099 /* if not a wild card name, check for an exact match */
7100 if ((nvl_keys
[k
].zkey_flags
& ZK_WILDCARDLIST
) == 0 &&
7101 strcmp(nvl_keys
[k
].zkey_name
, name
) != 0)
7104 identified
= B_TRUE
;
7106 if (nvl_keys
[k
].zkey_type
!= DATA_TYPE_ANY
&&
7107 nvl_keys
[k
].zkey_type
!= type
) {
7108 return (SET_ERROR(ZFS_ERR_IOC_ARG_BADTYPE
));
7111 if (nvl_keys
[k
].zkey_flags
& ZK_OPTIONAL
)
7114 required_keys_found
= B_TRUE
;
7118 /* allow an 'optional' key, everything else is invalid */
7120 (strcmp(name
, "optional") != 0 ||
7121 type
!= DATA_TYPE_NVLIST
)) {
7122 return (SET_ERROR(ZFS_ERR_IOC_ARG_UNAVAIL
));
7126 /* verify that all required keys were found */
7127 for (int k
= 0; k
< vec
->zvec_nvl_key_count
; k
++) {
7128 if (nvl_keys
[k
].zkey_flags
& ZK_OPTIONAL
)
7131 if (nvl_keys
[k
].zkey_flags
& ZK_WILDCARDLIST
) {
7132 /* at least one non-optional key is expected here */
7133 if (!required_keys_found
)
7134 return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED
));
7138 if (!nvlist_exists(innvl
, nvl_keys
[k
].zkey_name
))
7139 return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED
));
7146 pool_status_check(const char *name
, zfs_ioc_namecheck_t type
,
7147 zfs_ioc_poolcheck_t check
)
7152 ASSERT(type
== POOL_NAME
|| type
== DATASET_NAME
||
7153 type
== ENTITY_NAME
);
7155 if (check
& POOL_CHECK_NONE
)
7158 error
= spa_open(name
, &spa
, FTAG
);
7160 if ((check
& POOL_CHECK_SUSPENDED
) && spa_suspended(spa
))
7161 error
= SET_ERROR(EAGAIN
);
7162 else if ((check
& POOL_CHECK_READONLY
) && !spa_writeable(spa
))
7163 error
= SET_ERROR(EROFS
);
7164 spa_close(spa
, FTAG
);
7170 zfsdev_get_state_impl(minor_t minor
, enum zfsdev_state_type which
)
7174 for (zs
= zfsdev_state_list
; zs
!= NULL
; zs
= zs
->zs_next
) {
7175 if (zs
->zs_minor
== minor
) {
7179 return (zs
->zs_onexit
);
7181 return (zs
->zs_zevent
);
7192 zfsdev_get_state(minor_t minor
, enum zfsdev_state_type which
)
7196 ptr
= zfsdev_get_state_impl(minor
, which
);
7202 * Find a free minor number. The zfsdev_state_list is expected to
7203 * be short since it is only a list of currently open file handles.
7206 zfsdev_minor_alloc(void)
7208 static minor_t last_minor
= 0;
7211 ASSERT(MUTEX_HELD(&zfsdev_state_lock
));
7213 for (m
= last_minor
+ 1; m
!= last_minor
; m
++) {
7214 if (m
> ZFSDEV_MAX_MINOR
)
7216 if (zfsdev_get_state_impl(m
, ZST_ALL
) == NULL
) {
7226 zfsdev_ioctl_common(uint_t vecnum
, unsigned long arg
)
7229 int error
, cmd
, rc
, flag
= 0;
7230 const zfs_ioc_vec_t
*vec
;
7231 char *saved_poolname
= NULL
;
7232 nvlist_t
*innvl
= NULL
;
7233 fstrans_cookie_t cookie
;
7236 if (vecnum
>= sizeof (zfs_ioc_vec
) / sizeof (zfs_ioc_vec
[0]))
7237 return (-SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL
));
7238 vec
= &zfs_ioc_vec
[vecnum
];
7241 * The registered ioctl list may be sparse, verify that either
7242 * a normal or legacy handler are registered.
7244 if (vec
->zvec_func
== NULL
&& vec
->zvec_legacy_func
== NULL
)
7245 return (-SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL
));
7247 zc
= kmem_zalloc(sizeof (zfs_cmd_t
), KM_SLEEP
);
7249 error
= ddi_copyin((void *)(uintptr_t)arg
, zc
, sizeof (zfs_cmd_t
),
7252 error
= SET_ERROR(EFAULT
);
7256 zc
->zc_iflags
= flag
& FKIOCTL
;
7257 if (zc
->zc_nvlist_src_size
> MAX_NVLIST_SRC_SIZE
) {
7259 * Make sure the user doesn't pass in an insane value for
7260 * zc_nvlist_src_size. We have to check, since we will end
7261 * up allocating that much memory inside of get_nvlist(). This
7262 * prevents a nefarious user from allocating tons of kernel
7265 * Also, we return EINVAL instead of ENOMEM here. The reason
7266 * being that returning ENOMEM from an ioctl() has a special
7267 * connotation; that the user's size value is too small and
7268 * needs to be expanded to hold the nvlist. See
7269 * zcmd_expand_dst_nvlist() for details.
7271 error
= SET_ERROR(EINVAL
); /* User's size too big */
7273 } else if (zc
->zc_nvlist_src_size
!= 0) {
7274 error
= get_nvlist(zc
->zc_nvlist_src
, zc
->zc_nvlist_src_size
,
7275 zc
->zc_iflags
, &innvl
);
7281 * Ensure that all pool/dataset names are valid before we pass down to
7284 zc
->zc_name
[sizeof (zc
->zc_name
) - 1] = '\0';
7285 switch (vec
->zvec_namecheck
) {
7287 if (pool_namecheck(zc
->zc_name
, NULL
, NULL
) != 0)
7288 error
= SET_ERROR(EINVAL
);
7290 error
= pool_status_check(zc
->zc_name
,
7291 vec
->zvec_namecheck
, vec
->zvec_pool_check
);
7295 if (dataset_namecheck(zc
->zc_name
, NULL
, NULL
) != 0)
7296 error
= SET_ERROR(EINVAL
);
7298 error
= pool_status_check(zc
->zc_name
,
7299 vec
->zvec_namecheck
, vec
->zvec_pool_check
);
7303 if (entity_namecheck(zc
->zc_name
, NULL
, NULL
) != 0) {
7304 error
= SET_ERROR(EINVAL
);
7306 error
= pool_status_check(zc
->zc_name
,
7307 vec
->zvec_namecheck
, vec
->zvec_pool_check
);
7315 * Ensure that all input pairs are valid before we pass them down
7316 * to the lower layers.
7318 * The vectored functions can use fnvlist_lookup_{type} for any
7319 * required pairs since zfs_check_input_nvpairs() confirmed that
7320 * they exist and are of the correct type.
7322 if (error
== 0 && vec
->zvec_func
!= NULL
) {
7323 error
= zfs_check_input_nvpairs(innvl
, vec
);
7329 cookie
= spl_fstrans_mark();
7330 error
= vec
->zvec_secpolicy(zc
, innvl
, CRED());
7331 spl_fstrans_unmark(cookie
);
7337 /* legacy ioctls can modify zc_name */
7338 saved_poolname
= strdup(zc
->zc_name
);
7339 if (saved_poolname
== NULL
) {
7340 error
= SET_ERROR(ENOMEM
);
7343 saved_poolname
[strcspn(saved_poolname
, "/@#")] = '\0';
7346 if (vec
->zvec_func
!= NULL
) {
7350 nvlist_t
*lognv
= NULL
;
7352 ASSERT(vec
->zvec_legacy_func
== NULL
);
7355 * Add the innvl to the lognv before calling the func,
7356 * in case the func changes the innvl.
7358 if (vec
->zvec_allow_log
) {
7359 lognv
= fnvlist_alloc();
7360 fnvlist_add_string(lognv
, ZPOOL_HIST_IOCTL
,
7362 if (!nvlist_empty(innvl
)) {
7363 fnvlist_add_nvlist(lognv
, ZPOOL_HIST_INPUT_NVL
,
7368 outnvl
= fnvlist_alloc();
7369 cookie
= spl_fstrans_mark();
7370 error
= vec
->zvec_func(zc
->zc_name
, innvl
, outnvl
);
7371 spl_fstrans_unmark(cookie
);
7374 * Some commands can partially execute, modify state, and still
7375 * return an error. In these cases, attempt to record what
7379 (cmd
== ZFS_IOC_CHANNEL_PROGRAM
&& error
!= EINVAL
)) &&
7380 vec
->zvec_allow_log
&&
7381 spa_open(zc
->zc_name
, &spa
, FTAG
) == 0) {
7382 if (!nvlist_empty(outnvl
)) {
7383 fnvlist_add_nvlist(lognv
, ZPOOL_HIST_OUTPUT_NVL
,
7387 fnvlist_add_int64(lognv
, ZPOOL_HIST_ERRNO
,
7390 (void) spa_history_log_nvl(spa
, lognv
);
7391 spa_close(spa
, FTAG
);
7393 fnvlist_free(lognv
);
7395 if (!nvlist_empty(outnvl
) || zc
->zc_nvlist_dst_size
!= 0) {
7397 if (vec
->zvec_smush_outnvlist
) {
7398 smusherror
= nvlist_smush(outnvl
,
7399 zc
->zc_nvlist_dst_size
);
7401 if (smusherror
== 0)
7402 puterror
= put_nvlist(zc
, outnvl
);
7408 nvlist_free(outnvl
);
7410 cookie
= spl_fstrans_mark();
7411 error
= vec
->zvec_legacy_func(zc
);
7412 spl_fstrans_unmark(cookie
);
7417 rc
= ddi_copyout(zc
, (void *)(uintptr_t)arg
, sizeof (zfs_cmd_t
), flag
);
7418 if (error
== 0 && rc
!= 0)
7419 error
= SET_ERROR(EFAULT
);
7420 if (error
== 0 && vec
->zvec_allow_log
) {
7421 char *s
= tsd_get(zfs_allow_log_key
);
7424 (void) tsd_set(zfs_allow_log_key
, saved_poolname
);
7426 if (saved_poolname
!= NULL
)
7427 strfree(saved_poolname
);
7430 kmem_free(zc
, sizeof (zfs_cmd_t
));
7439 if ((error
= zvol_init()) != 0)
7442 spa_init(FREAD
| FWRITE
);
7447 mutex_init(&zfsdev_state_lock
, NULL
, MUTEX_DEFAULT
, NULL
);
7448 zfsdev_state_list
= kmem_zalloc(sizeof (zfsdev_state_t
), KM_SLEEP
);
7449 zfsdev_state_list
->zs_minor
= -1;
7451 if ((error
= zfsdev_attach()) != 0)
7454 tsd_create(&zfs_fsyncer_key
, NULL
);
7455 tsd_create(&rrw_tsd_key
, rrw_tsd_destroy
);
7456 tsd_create(&zfs_allow_log_key
, zfs_allow_log_destroy
);
7470 zfsdev_state_t
*zs
, *zsprev
= NULL
;
7474 mutex_destroy(&zfsdev_state_lock
);
7476 for (zs
= zfsdev_state_list
; zs
!= NULL
; zs
= zs
->zs_next
) {
7478 kmem_free(zsprev
, sizeof (zfsdev_state_t
));
7482 kmem_free(zsprev
, sizeof (zfsdev_state_t
));
7488 tsd_destroy(&zfs_fsyncer_key
);
7489 tsd_destroy(&rrw_tsd_key
);
7490 tsd_destroy(&zfs_allow_log_key
);