config/dracut/90zfs: handle cases where hostid(1) returns all zeros
[zfs.git] / include / sys / zfs_ioctl.h
blobafae576ea21a0b2b8ddfca169a82f5e31b7d3df6
1 /*
2 * CDDL HEADER START
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]
19 * CDDL HEADER END
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2012, 2020 by Delphix. All rights reserved.
24 * Copyright 2016 RackTop Systems.
25 * Copyright (c) 2017, Intel Corporation.
28 #ifndef _SYS_ZFS_IOCTL_H
29 #define _SYS_ZFS_IOCTL_H
31 #include <sys/cred.h>
32 #include <sys/dmu.h>
33 #include <sys/zio.h>
34 #include <sys/dsl_deleg.h>
35 #include <sys/spa.h>
36 #include <sys/zfs_stat.h>
38 #ifdef _KERNEL
39 #include <sys/nvpair.h>
40 #endif /* _KERNEL */
42 #ifdef __cplusplus
43 extern "C" {
44 #endif
47 * The structures in this file are passed between userland and the
48 * kernel. Userland may be running a 32-bit process, while the kernel
49 * is 64-bit. Therefore, these structures need to compile the same in
50 * 32-bit and 64-bit. This means not using type "long", and adding
51 * explicit padding so that the 32-bit structure will not be packed more
52 * tightly than the 64-bit structure (which requires 64-bit alignment).
56 * Property values for snapdir
58 #define ZFS_SNAPDIR_HIDDEN 0
59 #define ZFS_SNAPDIR_VISIBLE 1
62 * Property values for snapdev
64 #define ZFS_SNAPDEV_HIDDEN 0
65 #define ZFS_SNAPDEV_VISIBLE 1
67 * Property values for acltype
69 #define ZFS_ACLTYPE_OFF 0
70 #define ZFS_ACLTYPE_POSIX 1
71 #define ZFS_ACLTYPE_NFSV4 2
74 * Field manipulation macros for the drr_versioninfo field of the
75 * send stream header.
79 * Header types for zfs send streams.
81 typedef enum drr_headertype {
82 DMU_SUBSTREAM = 0x1,
83 DMU_COMPOUNDSTREAM = 0x2
84 } drr_headertype_t;
86 #define DMU_GET_STREAM_HDRTYPE(vi) BF64_GET((vi), 0, 2)
87 #define DMU_SET_STREAM_HDRTYPE(vi, x) BF64_SET((vi), 0, 2, x)
89 #define DMU_GET_FEATUREFLAGS(vi) BF64_GET((vi), 2, 30)
90 #define DMU_SET_FEATUREFLAGS(vi, x) BF64_SET((vi), 2, 30, x)
93 * Feature flags for zfs send streams (flags in drr_versioninfo)
96 #define DMU_BACKUP_FEATURE_DEDUP (1 << 0)
97 #define DMU_BACKUP_FEATURE_DEDUPPROPS (1 << 1)
98 #define DMU_BACKUP_FEATURE_SA_SPILL (1 << 2)
99 /* flags #3 - #15 are reserved for incompatible closed-source implementations */
100 #define DMU_BACKUP_FEATURE_EMBED_DATA (1 << 16)
101 #define DMU_BACKUP_FEATURE_LZ4 (1 << 17)
102 /* flag #18 is reserved for a Delphix feature */
103 #define DMU_BACKUP_FEATURE_LARGE_BLOCKS (1 << 19)
104 #define DMU_BACKUP_FEATURE_RESUMING (1 << 20)
105 #define DMU_BACKUP_FEATURE_REDACTED (1 << 21)
106 #define DMU_BACKUP_FEATURE_COMPRESSED (1 << 22)
107 #define DMU_BACKUP_FEATURE_LARGE_DNODE (1 << 23)
108 #define DMU_BACKUP_FEATURE_RAW (1 << 24)
109 #define DMU_BACKUP_FEATURE_ZSTD (1 << 25)
110 #define DMU_BACKUP_FEATURE_HOLDS (1 << 26)
112 * The SWITCH_TO_LARGE_BLOCKS feature indicates that we can receive
113 * incremental LARGE_BLOCKS streams (those with WRITE records of >128KB) even
114 * if the previous send did not use LARGE_BLOCKS, and thus its large blocks
115 * were split into multiple 128KB WRITE records. (See
116 * flush_write_batch_impl() and receive_object()). Older software that does
117 * not support this flag may encounter a bug when switching to large blocks,
118 * which causes files to incorrectly be zeroed.
120 * This flag is currently not set on any send streams. In the future, we
121 * intend for incremental send streams of snapshots that have large blocks to
122 * use LARGE_BLOCKS by default, and these streams will also have the
123 * SWITCH_TO_LARGE_BLOCKS feature set. This ensures that streams from the
124 * default use of "zfs send" won't encounter the bug mentioned above.
126 #define DMU_BACKUP_FEATURE_SWITCH_TO_LARGE_BLOCKS (1 << 27)
129 * Mask of all supported backup features
131 #define DMU_BACKUP_FEATURE_MASK (DMU_BACKUP_FEATURE_SA_SPILL | \
132 DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_LZ4 | \
133 DMU_BACKUP_FEATURE_RESUMING | DMU_BACKUP_FEATURE_LARGE_BLOCKS | \
134 DMU_BACKUP_FEATURE_COMPRESSED | DMU_BACKUP_FEATURE_LARGE_DNODE | \
135 DMU_BACKUP_FEATURE_RAW | DMU_BACKUP_FEATURE_HOLDS | \
136 DMU_BACKUP_FEATURE_REDACTED | DMU_BACKUP_FEATURE_SWITCH_TO_LARGE_BLOCKS | \
137 DMU_BACKUP_FEATURE_ZSTD)
139 /* Are all features in the given flag word currently supported? */
140 #define DMU_STREAM_SUPPORTED(x) (!((x) & ~DMU_BACKUP_FEATURE_MASK))
142 typedef enum dmu_send_resume_token_version {
143 ZFS_SEND_RESUME_TOKEN_VERSION = 1
144 } dmu_send_resume_token_version_t;
147 * The drr_versioninfo field of the dmu_replay_record has the
148 * following layout:
150 * 64 56 48 40 32 24 16 8 0
151 * +-------+-------+-------+-------+-------+-------+-------+-------+
152 * | reserved | feature-flags |C|S|
153 * +-------+-------+-------+-------+-------+-------+-------+-------+
155 * The low order two bits indicate the header type: SUBSTREAM (0x1)
156 * or COMPOUNDSTREAM (0x2). Using two bits for this is historical:
157 * this field used to be a version number, where the two version types
158 * were 1 and 2. Using two bits for this allows earlier versions of
159 * the code to be able to recognize send streams that don't use any
160 * of the features indicated by feature flags.
163 #define DMU_BACKUP_MAGIC 0x2F5bacbacULL
166 * Send stream flags. Bits 24-31 are reserved for vendor-specific
167 * implementations and should not be used.
169 #define DRR_FLAG_CLONE (1<<0)
170 #define DRR_FLAG_CI_DATA (1<<1)
172 * This send stream, if it is a full send, includes the FREE and FREEOBJECT
173 * records that are created by the sending process. This means that the send
174 * stream can be received as a clone, even though it is not an incremental.
175 * This is not implemented as a feature flag, because the receiving side does
176 * not need to have implemented it to receive this stream; it is fully backwards
177 * compatible. We need a flag, though, because full send streams without it
178 * cannot necessarily be received as a clone correctly.
180 #define DRR_FLAG_FREERECORDS (1<<2)
182 * When DRR_FLAG_SPILL_BLOCK is set it indicates the DRR_OBJECT_SPILL
183 * and DRR_SPILL_UNMODIFIED flags are meaningful in the send stream.
185 * When DRR_FLAG_SPILL_BLOCK is set, DRR_OBJECT records will have
186 * DRR_OBJECT_SPILL set if and only if they should have a spill block
187 * (either an existing one, or a new one in the send stream). When clear
188 * the object does not have a spill block and any existing spill block
189 * should be freed.
191 * Similarly, when DRR_FLAG_SPILL_BLOCK is set, DRR_SPILL records will
192 * have DRR_SPILL_UNMODIFIED set if and only if they were included for
193 * backward compatibility purposes, and can be safely ignored by new versions
194 * of zfs receive. Previous versions of ZFS which do not understand the
195 * DRR_FLAG_SPILL_BLOCK will process this record and recreate any missing
196 * spill blocks.
198 #define DRR_FLAG_SPILL_BLOCK (1<<3)
201 * flags in the drr_flags field in the DRR_WRITE, DRR_SPILL, DRR_OBJECT,
202 * DRR_WRITE_BYREF, and DRR_OBJECT_RANGE blocks
204 #define DRR_CHECKSUM_DEDUP (1<<0) /* not used for SPILL records */
205 #define DRR_RAW_BYTESWAP (1<<1)
206 #define DRR_OBJECT_SPILL (1<<2) /* OBJECT record has a spill block */
207 #define DRR_SPILL_UNMODIFIED (1<<2) /* SPILL record for unmodified block */
209 #define DRR_IS_DEDUP_CAPABLE(flags) ((flags) & DRR_CHECKSUM_DEDUP)
210 #define DRR_IS_RAW_BYTESWAPPED(flags) ((flags) & DRR_RAW_BYTESWAP)
211 #define DRR_OBJECT_HAS_SPILL(flags) ((flags) & DRR_OBJECT_SPILL)
212 #define DRR_SPILL_IS_UNMODIFIED(flags) ((flags) & DRR_SPILL_UNMODIFIED)
214 /* deal with compressed drr_write replay records */
215 #define DRR_WRITE_COMPRESSED(drrw) ((drrw)->drr_compressiontype != 0)
216 #define DRR_WRITE_PAYLOAD_SIZE(drrw) \
217 (DRR_WRITE_COMPRESSED(drrw) ? (drrw)->drr_compressed_size : \
218 (drrw)->drr_logical_size)
219 #define DRR_SPILL_PAYLOAD_SIZE(drrs) \
220 ((drrs)->drr_compressed_size ? \
221 (drrs)->drr_compressed_size : (drrs)->drr_length)
222 #define DRR_OBJECT_PAYLOAD_SIZE(drro) \
223 ((drro)->drr_raw_bonuslen != 0 ? \
224 (drro)->drr_raw_bonuslen : P2ROUNDUP((drro)->drr_bonuslen, 8))
227 * zfs ioctl command structure
230 /* Header is used in C++ so can't forward declare untagged struct */
231 struct drr_begin {
232 uint64_t drr_magic;
233 uint64_t drr_versioninfo; /* was drr_version */
234 uint64_t drr_creation_time;
235 dmu_objset_type_t drr_type;
236 uint32_t drr_flags;
237 uint64_t drr_toguid;
238 uint64_t drr_fromguid;
239 char drr_toname[MAXNAMELEN];
242 typedef struct dmu_replay_record {
243 enum {
244 DRR_BEGIN, DRR_OBJECT, DRR_FREEOBJECTS,
245 DRR_WRITE, DRR_FREE, DRR_END, DRR_WRITE_BYREF,
246 DRR_SPILL, DRR_WRITE_EMBEDDED, DRR_OBJECT_RANGE, DRR_REDACT,
247 DRR_NUMTYPES
248 } drr_type;
249 uint32_t drr_payloadlen;
250 union {
251 struct drr_begin drr_begin;
252 struct drr_end {
253 zio_cksum_t drr_checksum;
254 uint64_t drr_toguid;
255 } drr_end;
256 struct drr_object {
257 uint64_t drr_object;
258 dmu_object_type_t drr_type;
259 dmu_object_type_t drr_bonustype;
260 uint32_t drr_blksz;
261 uint32_t drr_bonuslen;
262 uint8_t drr_checksumtype;
263 uint8_t drr_compress;
264 uint8_t drr_dn_slots;
265 uint8_t drr_flags;
266 uint32_t drr_raw_bonuslen;
267 uint64_t drr_toguid;
268 /* only (possibly) nonzero for raw streams */
269 uint8_t drr_indblkshift;
270 uint8_t drr_nlevels;
271 uint8_t drr_nblkptr;
272 uint8_t drr_pad[5];
273 uint64_t drr_maxblkid;
274 /* bonus content follows */
275 } drr_object;
276 struct drr_freeobjects {
277 uint64_t drr_firstobj;
278 uint64_t drr_numobjs;
279 uint64_t drr_toguid;
280 } drr_freeobjects;
281 struct drr_write {
282 uint64_t drr_object;
283 dmu_object_type_t drr_type;
284 uint32_t drr_pad;
285 uint64_t drr_offset;
286 uint64_t drr_logical_size;
287 uint64_t drr_toguid;
288 uint8_t drr_checksumtype;
289 uint8_t drr_flags;
290 uint8_t drr_compressiontype;
291 uint8_t drr_pad2[5];
292 /* deduplication key */
293 ddt_key_t drr_key;
294 /* only nonzero if drr_compressiontype is not 0 */
295 uint64_t drr_compressed_size;
296 /* only nonzero for raw streams */
297 uint8_t drr_salt[ZIO_DATA_SALT_LEN];
298 uint8_t drr_iv[ZIO_DATA_IV_LEN];
299 uint8_t drr_mac[ZIO_DATA_MAC_LEN];
300 /* content follows */
301 } drr_write;
302 struct drr_free {
303 uint64_t drr_object;
304 uint64_t drr_offset;
305 uint64_t drr_length;
306 uint64_t drr_toguid;
307 } drr_free;
308 struct drr_write_byref {
309 /* where to put the data */
310 uint64_t drr_object;
311 uint64_t drr_offset;
312 uint64_t drr_length;
313 uint64_t drr_toguid;
314 /* where to find the prior copy of the data */
315 uint64_t drr_refguid;
316 uint64_t drr_refobject;
317 uint64_t drr_refoffset;
318 /* properties of the data */
319 uint8_t drr_checksumtype;
320 uint8_t drr_flags;
321 uint8_t drr_pad2[6];
322 ddt_key_t drr_key; /* deduplication key */
323 } drr_write_byref;
324 struct drr_spill {
325 uint64_t drr_object;
326 uint64_t drr_length;
327 uint64_t drr_toguid;
328 uint8_t drr_flags;
329 uint8_t drr_compressiontype;
330 uint8_t drr_pad[6];
331 /* only nonzero for raw streams */
332 uint64_t drr_compressed_size;
333 uint8_t drr_salt[ZIO_DATA_SALT_LEN];
334 uint8_t drr_iv[ZIO_DATA_IV_LEN];
335 uint8_t drr_mac[ZIO_DATA_MAC_LEN];
336 dmu_object_type_t drr_type;
337 /* spill data follows */
338 } drr_spill;
339 struct drr_write_embedded {
340 uint64_t drr_object;
341 uint64_t drr_offset;
342 /* logical length, should equal blocksize */
343 uint64_t drr_length;
344 uint64_t drr_toguid;
345 uint8_t drr_compression;
346 uint8_t drr_etype;
347 uint8_t drr_pad[6];
348 uint32_t drr_lsize; /* uncompressed size of payload */
349 uint32_t drr_psize; /* compr. (real) size of payload */
350 /* (possibly compressed) content follows */
351 } drr_write_embedded;
352 struct drr_object_range {
353 uint64_t drr_firstobj;
354 uint64_t drr_numslots;
355 uint64_t drr_toguid;
356 uint8_t drr_salt[ZIO_DATA_SALT_LEN];
357 uint8_t drr_iv[ZIO_DATA_IV_LEN];
358 uint8_t drr_mac[ZIO_DATA_MAC_LEN];
359 uint8_t drr_flags;
360 uint8_t drr_pad[3];
361 } drr_object_range;
362 struct drr_redact {
363 uint64_t drr_object;
364 uint64_t drr_offset;
365 uint64_t drr_length;
366 uint64_t drr_toguid;
367 } drr_redact;
370 * Note: drr_checksum is overlaid with all record types
371 * except DRR_BEGIN. Therefore its (non-pad) members
372 * must not overlap with members from the other structs.
373 * We accomplish this by putting its members at the very
374 * end of the struct.
376 struct drr_checksum {
377 uint64_t drr_pad[34];
379 * fletcher-4 checksum of everything preceding the
380 * checksum.
382 zio_cksum_t drr_checksum;
383 } drr_checksum;
384 } drr_u;
385 } dmu_replay_record_t;
387 /* diff record range types */
388 typedef enum diff_type {
389 DDR_NONE = 0x1,
390 DDR_INUSE = 0x2,
391 DDR_FREE = 0x4
392 } diff_type_t;
395 * The diff reports back ranges of free or in-use objects.
397 typedef struct dmu_diff_record {
398 uint64_t ddr_type;
399 uint64_t ddr_first;
400 uint64_t ddr_last;
401 } dmu_diff_record_t;
403 typedef struct zinject_record {
404 uint64_t zi_objset;
405 uint64_t zi_object;
406 uint64_t zi_start;
407 uint64_t zi_end;
408 uint64_t zi_guid;
409 uint32_t zi_level;
410 uint32_t zi_error;
411 uint64_t zi_type;
412 uint32_t zi_freq;
413 uint32_t zi_failfast;
414 char zi_func[MAXNAMELEN];
415 uint32_t zi_iotype;
416 int32_t zi_duration;
417 uint64_t zi_timer;
418 uint64_t zi_nlanes;
419 uint32_t zi_cmd;
420 uint32_t zi_dvas;
421 } zinject_record_t;
423 #define ZINJECT_NULL 0x1
424 #define ZINJECT_FLUSH_ARC 0x2
425 #define ZINJECT_UNLOAD_SPA 0x4
426 #define ZINJECT_CALC_RANGE 0x8
428 #define ZEVENT_NONE 0x0
429 #define ZEVENT_NONBLOCK 0x1
430 #define ZEVENT_SIZE 1024
432 #define ZEVENT_SEEK_START 0
433 #define ZEVENT_SEEK_END UINT64_MAX
435 /* scaled frequency ranges */
436 #define ZI_PERCENTAGE_MIN 4294UL
437 #define ZI_PERCENTAGE_MAX UINT32_MAX
439 #define ZI_NO_DVA (-1)
441 typedef enum zinject_type {
442 ZINJECT_UNINITIALIZED,
443 ZINJECT_DATA_FAULT,
444 ZINJECT_DEVICE_FAULT,
445 ZINJECT_LABEL_FAULT,
446 ZINJECT_IGNORED_WRITES,
447 ZINJECT_PANIC,
448 ZINJECT_DELAY_IO,
449 ZINJECT_DECRYPT_FAULT,
450 } zinject_type_t;
452 typedef struct zfs_share {
453 uint64_t z_exportdata;
454 uint64_t z_sharedata;
455 uint64_t z_sharetype; /* 0 = share, 1 = unshare */
456 uint64_t z_sharemax; /* max length of share string */
457 } zfs_share_t;
460 * ZFS file systems may behave the usual, POSIX-compliant way, where
461 * name lookups are case-sensitive. They may also be set up so that
462 * all the name lookups are case-insensitive, or so that only some
463 * lookups, the ones that set an FIGNORECASE flag, are case-insensitive.
465 typedef enum zfs_case {
466 ZFS_CASE_SENSITIVE,
467 ZFS_CASE_INSENSITIVE,
468 ZFS_CASE_MIXED
469 } zfs_case_t;
472 * Note: this struct must have the same layout in 32-bit and 64-bit, so
473 * that 32-bit processes (like /sbin/zfs) can pass it to the 64-bit
474 * kernel. Therefore, we add padding to it so that no "hidden" padding
475 * is automatically added on 64-bit (but not on 32-bit).
477 typedef struct zfs_cmd {
478 char zc_name[MAXPATHLEN]; /* name of pool or dataset */
479 uint64_t zc_nvlist_src; /* really (char *) */
480 uint64_t zc_nvlist_src_size;
481 uint64_t zc_nvlist_dst; /* really (char *) */
482 uint64_t zc_nvlist_dst_size;
483 boolean_t zc_nvlist_dst_filled; /* put an nvlist in dst? */
484 int zc_pad2;
487 * The following members are for legacy ioctls which haven't been
488 * converted to the new method.
490 uint64_t zc_history; /* really (char *) */
491 char zc_value[MAXPATHLEN * 2];
492 char zc_string[MAXNAMELEN];
493 uint64_t zc_guid;
494 uint64_t zc_nvlist_conf; /* really (char *) */
495 uint64_t zc_nvlist_conf_size;
496 uint64_t zc_cookie;
497 uint64_t zc_objset_type;
498 uint64_t zc_perm_action;
499 uint64_t zc_history_len;
500 uint64_t zc_history_offset;
501 uint64_t zc_obj;
502 uint64_t zc_iflags; /* internal to zfs(7fs) */
503 zfs_share_t zc_share;
504 dmu_objset_stats_t zc_objset_stats;
505 struct drr_begin zc_begin_record;
506 zinject_record_t zc_inject_record;
507 uint32_t zc_defer_destroy;
508 uint32_t zc_flags;
509 uint64_t zc_action_handle;
510 int zc_cleanup_fd;
511 uint8_t zc_simple;
512 uint8_t zc_pad[3]; /* alignment */
513 uint64_t zc_sendobj;
514 uint64_t zc_fromobj;
515 uint64_t zc_createtxg;
516 zfs_stat_t zc_stat;
517 uint64_t zc_zoneid;
518 } zfs_cmd_t;
520 typedef struct zfs_useracct {
521 char zu_domain[256];
522 uid_t zu_rid;
523 uint32_t zu_pad;
524 uint64_t zu_space;
525 } zfs_useracct_t;
527 #define ZFSDEV_MAX_MINOR (1 << 16)
528 #define ZFS_MIN_MINOR (ZFSDEV_MAX_MINOR + 1)
530 #define ZPOOL_EXPORT_AFTER_SPLIT 0x1
532 #ifdef _KERNEL
533 struct objset;
534 struct zfsvfs;
536 typedef struct zfs_creat {
537 nvlist_t *zct_zplprops;
538 nvlist_t *zct_props;
539 } zfs_creat_t;
541 extern int zfs_secpolicy_snapshot_perms(const char *, cred_t *);
542 extern int zfs_secpolicy_rename_perms(const char *, const char *, cred_t *);
543 extern int zfs_secpolicy_destroy_perms(const char *, cred_t *);
544 extern void zfs_unmount_snap(const char *);
545 extern void zfs_destroy_unmount_origin(const char *);
546 extern int getzfsvfs_impl(struct objset *, struct zfsvfs **);
547 extern int getzfsvfs(const char *, struct zfsvfs **);
549 enum zfsdev_state_type {
550 ZST_ONEXIT,
551 ZST_ZEVENT,
552 ZST_ALL,
556 * The zfsdev_state_t structure is managed as a singly-linked list
557 * from which items are never deleted. This allows for lock-free
558 * reading of the list so long as assignments to the zs_next and
559 * reads from zs_minor are performed atomically. Empty items are
560 * indicated by storing -1 into zs_minor.
562 typedef struct zfsdev_state {
563 struct zfsdev_state *zs_next; /* next zfsdev_state_t link */
564 minor_t zs_minor; /* made up minor number */
565 void *zs_onexit; /* onexit data */
566 void *zs_zevent; /* zevent data */
567 } zfsdev_state_t;
569 extern void *zfsdev_get_state(minor_t minor, enum zfsdev_state_type which);
570 extern int zfsdev_getminor(int fd, minor_t *minorp);
571 extern minor_t zfsdev_minor_alloc(void);
573 extern uint_t zfs_fsyncer_key;
574 extern uint_t zfs_allow_log_key;
576 #endif /* _KERNEL */
578 #ifdef __cplusplus
580 #endif
582 #endif /* _SYS_ZFS_IOCTL_H */