1/*
2 * CDDL HEADER START
3 *
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.
7 *
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.
12 *
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]
18 *
19 * CDDL HEADER END
20 */
21/*
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2011-2012 Pawel Jakub Dawidek. All rights reserved.
24 * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
25 * Copyright 2016 RackTop Systems.
26 * Copyright (c) 2014 Integros [integros.com]
27 * Copyright (c) 2017, Intel Corporation.
28 */
29
30#ifndef	_SYS_ZFS_IOCTL_H
31#define	_SYS_ZFS_IOCTL_H
32
33#include <sys/cred.h>
34#include <sys/dmu.h>
35#include <sys/zio.h>
36#include <sys/dsl_deleg.h>
37#include <sys/spa.h>
38#include <sys/zfs_stat.h>
39
40#ifdef _KERNEL
41#include <sys/nvpair.h>
42#endif	/* _KERNEL */
43
44#ifdef	__cplusplus
45extern "C" {
46#endif
47
48/*
49 * The structures in this file are passed between userland and the
50 * kernel.  Userland may be running a 32-bit process, while the kernel
51 * is 64-bit.  Therefore, these structures need to compile the same in
52 * 32-bit and 64-bit.  This means not using type "long", and adding
53 * explicit padding so that the 32-bit structure will not be packed more
54 * tightly than the 64-bit structure (which requires 64-bit alignment).
55 */
56
57/*
58 * Property values for snapdir
59 */
60#define	ZFS_SNAPDIR_HIDDEN		0
61#define	ZFS_SNAPDIR_VISIBLE		1
62
63/*
64 * Field manipulation macros for the drr_versioninfo field of the
65 * send stream header.
66 */
67
68/*
69 * Header types for zfs send streams.
70 */
71typedef enum drr_headertype {
72	DMU_SUBSTREAM = 0x1,
73	DMU_COMPOUNDSTREAM = 0x2
74} drr_headertype_t;
75
76#define	DMU_GET_STREAM_HDRTYPE(vi)	BF64_GET((vi), 0, 2)
77#define	DMU_SET_STREAM_HDRTYPE(vi, x)	BF64_SET((vi), 0, 2, x)
78
79#define	DMU_GET_FEATUREFLAGS(vi)	BF64_GET((vi), 2, 30)
80#define	DMU_SET_FEATUREFLAGS(vi, x)	BF64_SET((vi), 2, 30, x)
81
82/*
83 * Feature flags for zfs send streams (flags in drr_versioninfo)
84 */
85
86#define	DMU_BACKUP_FEATURE_DEDUP		(1 << 0)
87#define	DMU_BACKUP_FEATURE_DEDUPPROPS		(1 << 1)
88#define	DMU_BACKUP_FEATURE_SA_SPILL		(1 << 2)
89/* flags #3 - #15 are reserved for incompatible closed-source implementations */
90#define	DMU_BACKUP_FEATURE_EMBED_DATA		(1 << 16)
91#define	DMU_BACKUP_FEATURE_LZ4			(1 << 17)
92/* flag #18 is reserved for a Delphix feature */
93#define	DMU_BACKUP_FEATURE_LARGE_BLOCKS		(1 << 19)
94#define	DMU_BACKUP_FEATURE_RESUMING		(1 << 20)
95/* flag #21 is reserved for the redacted send/receive feature */
96#define	DMU_BACKUP_FEATURE_COMPRESSED		(1 << 22)
97#define	DMU_BACKUP_FEATURE_LARGE_DNODE		(1 << 23)
98#define	DMU_BACKUP_FEATURE_RAW			(1 << 24)
99/* flag #25 is reserved for the ZSTD compression feature */
100#define	DMU_BACKUP_FEATURE_HOLDS		(1 << 26)
101
102/*
103 * Mask of all supported backup features
104 */
105#define	DMU_BACKUP_FEATURE_MASK	(DMU_BACKUP_FEATURE_DEDUP | \
106    DMU_BACKUP_FEATURE_DEDUPPROPS | DMU_BACKUP_FEATURE_SA_SPILL | \
107    DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_LZ4 | \
108    DMU_BACKUP_FEATURE_RESUMING | DMU_BACKUP_FEATURE_LARGE_BLOCKS | \
109    DMU_BACKUP_FEATURE_COMPRESSED | DMU_BACKUP_FEATURE_LARGE_DNODE | \
110    DMU_BACKUP_FEATURE_RAW | DMU_BACKUP_FEATURE_HOLDS)
111
112/* Are all features in the given flag word currently supported? */
113#define	DMU_STREAM_SUPPORTED(x)	(!((x) & ~DMU_BACKUP_FEATURE_MASK))
114
115typedef enum dmu_send_resume_token_version {
116	ZFS_SEND_RESUME_TOKEN_VERSION = 1
117} dmu_send_resume_token_version_t;
118
119/*
120 * The drr_versioninfo field of the dmu_replay_record has the
121 * following layout:
122 *
123 *	64	56	48	40	32	24	16	8	0
124 *	+-------+-------+-------+-------+-------+-------+-------+-------+
125 *	|		reserved	|        feature-flags	    |C|S|
126 *	+-------+-------+-------+-------+-------+-------+-------+-------+
127 *
128 * The low order two bits indicate the header type: SUBSTREAM (0x1)
129 * or COMPOUNDSTREAM (0x2).  Using two bits for this is historical:
130 * this field used to be a version number, where the two version types
131 * were 1 and 2.  Using two bits for this allows earlier versions of
132 * the code to be able to recognize send streams that don't use any
133 * of the features indicated by feature flags.
134 */
135
136#define	DMU_BACKUP_MAGIC 0x2F5bacbacULL
137
138/*
139 * Send stream flags.  Bits 24-31 are reserved for vendor-specific
140 * implementations and should not be used.
141 */
142#define	DRR_FLAG_CLONE		(1<<0)
143#define	DRR_FLAG_CI_DATA	(1<<1)
144/*
145 * This send stream, if it is a full send, includes the FREE and FREEOBJECT
146 * records that are created by the sending process.  This means that the send
147 * stream can be received as a clone, even though it is not an incremental.
148 * This is not implemented as a feature flag, because the receiving side does
149 * not need to have implemented it to receive this stream; it is fully backwards
150 * compatible.  We need a flag, though, because full send streams without it
151 * cannot necessarily be received as a clone correctly.
152 */
153#define	DRR_FLAG_FREERECORDS	(1<<2)
154/*
155 * When DRR_FLAG_SPILL_BLOCK is set it indicates the DRR_OBJECT_SPILL
156 * and DRR_SPILL_UNMODIFIED flags are meaningful in the send stream.
157 *
158 * When DRR_FLAG_SPILL_BLOCK is set, DRR_OBJECT records will have
159 * DRR_OBJECT_SPILL set if and only if they should have a spill block
160 * (either an existing one, or a new one in the send stream).  When clear
161 * the object does not have a spill block and any existing spill block
162 * should be freed.
163 *
164 * Similarly, when DRR_FLAG_SPILL_BLOCK is set, DRR_SPILL records will
165 * have DRR_SPILL_UNMODIFIED set if and only if they were included for
166 * backward compatibility purposes, and can be safely ignored by new versions
167 * of zfs receive.  Previous versions of ZFS which do not understand the
168 * DRR_FLAG_SPILL_BLOCK will process this record and recreate any missing
169 * spill blocks.
170 */
171#define	DRR_FLAG_SPILL_BLOCK	(1<<3)
172
173/*
174 * flags in the drr_flags field in the DRR_WRITE, DRR_SPILL, DRR_OBJECT,
175 * DRR_WRITE_BYREF, and DRR_OBJECT_RANGE blocks
176 */
177#define	DRR_CHECKSUM_DEDUP	(1<<0) /* not used for SPILL records */
178#define	DRR_RAW_BYTESWAP	(1<<1)
179#define	DRR_OBJECT_SPILL	(1<<2) /* OBJECT record has a spill block */
180#define	DRR_SPILL_UNMODIFIED	(1<<2) /* SPILL record for unmodified block */
181
182#define	DRR_IS_DEDUP_CAPABLE(flags)	((flags) & DRR_CHECKSUM_DEDUP)
183#define	DRR_IS_RAW_BYTESWAPPED(flags)	((flags) & DRR_RAW_BYTESWAP)
184#define	DRR_OBJECT_HAS_SPILL(flags)	((flags) & DRR_OBJECT_SPILL)
185#define	DRR_SPILL_IS_UNMODIFIED(flags)	((flags) & DRR_SPILL_UNMODIFIED)
186
187/* deal with compressed drr_write replay records */
188#define	DRR_WRITE_COMPRESSED(drrw)	((drrw)->drr_compressiontype != 0)
189#define	DRR_WRITE_PAYLOAD_SIZE(drrw) \
190	(DRR_WRITE_COMPRESSED(drrw) ? (drrw)->drr_compressed_size : \
191	(drrw)->drr_logical_size)
192#define	DRR_SPILL_PAYLOAD_SIZE(drrs) \
193	((drrs)->drr_compressed_size ? \
194	(drrs)->drr_compressed_size : (drrs)->drr_length)
195#define	DRR_OBJECT_PAYLOAD_SIZE(drro) \
196	((drro)->drr_raw_bonuslen != 0 ? \
197	(drro)->drr_raw_bonuslen : P2ROUNDUP((drro)->drr_bonuslen, 8))
198
199/*
200 * zfs ioctl command structure
201 */
202typedef struct dmu_replay_record {
203	enum {
204		DRR_BEGIN, DRR_OBJECT, DRR_FREEOBJECTS,
205		DRR_WRITE, DRR_FREE, DRR_END, DRR_WRITE_BYREF,
206		DRR_SPILL, DRR_WRITE_EMBEDDED, DRR_OBJECT_RANGE,
207		DRR_NUMTYPES
208	} drr_type;
209	uint32_t drr_payloadlen;
210	union {
211		struct drr_begin {
212			uint64_t drr_magic;
213			uint64_t drr_versioninfo; /* was drr_version */
214			uint64_t drr_creation_time;
215			dmu_objset_type_t drr_type;
216			uint32_t drr_flags;
217			uint64_t drr_toguid;
218			uint64_t drr_fromguid;
219			char drr_toname[MAXNAMELEN];
220		} drr_begin;
221		struct drr_end {
222			zio_cksum_t drr_checksum;
223			uint64_t drr_toguid;
224		} drr_end;
225		struct drr_object {
226			uint64_t drr_object;
227			dmu_object_type_t drr_type;
228			dmu_object_type_t drr_bonustype;
229			uint32_t drr_blksz;
230			uint32_t drr_bonuslen;
231			uint8_t drr_checksumtype;
232			uint8_t drr_compress;
233			uint8_t drr_dn_slots;
234			uint8_t drr_flags;
235			uint32_t drr_raw_bonuslen;
236			uint64_t drr_toguid;
237			/* only (possibly) nonzero for raw streams */
238			uint8_t drr_indblkshift;
239			uint8_t drr_nlevels;
240			uint8_t drr_nblkptr;
241			uint8_t drr_pad[5];
242			uint64_t drr_maxblkid;
243			/* bonus content follows */
244		} drr_object;
245		struct drr_freeobjects {
246			uint64_t drr_firstobj;
247			uint64_t drr_numobjs;
248			uint64_t drr_toguid;
249		} drr_freeobjects;
250		struct drr_write {
251			uint64_t drr_object;
252			dmu_object_type_t drr_type;
253			uint32_t drr_pad;
254			uint64_t drr_offset;
255			uint64_t drr_logical_size;
256			uint64_t drr_toguid;
257			uint8_t drr_checksumtype;
258			uint8_t drr_flags;
259			uint8_t drr_compressiontype;
260			uint8_t drr_pad2[5];
261			/* deduplication key */
262			ddt_key_t drr_key;
263			/* only nonzero if drr_compressiontype is not 0 */
264			uint64_t drr_compressed_size;
265			/* only nonzero for raw streams */
266			uint8_t drr_salt[ZIO_DATA_SALT_LEN];
267			uint8_t drr_iv[ZIO_DATA_IV_LEN];
268			uint8_t drr_mac[ZIO_DATA_MAC_LEN];
269			/* content follows */
270		} drr_write;
271		struct drr_free {
272			uint64_t drr_object;
273			uint64_t drr_offset;
274			uint64_t drr_length;
275			uint64_t drr_toguid;
276		} drr_free;
277		struct drr_write_byref {
278			/* where to put the data */
279			uint64_t drr_object;
280			uint64_t drr_offset;
281			uint64_t drr_length;
282			uint64_t drr_toguid;
283			/* where to find the prior copy of the data */
284			uint64_t drr_refguid;
285			uint64_t drr_refobject;
286			uint64_t drr_refoffset;
287			/* properties of the data */
288			uint8_t drr_checksumtype;
289			uint8_t drr_flags;
290			uint8_t drr_pad2[6];
291			ddt_key_t drr_key; /* deduplication key */
292		} drr_write_byref;
293		struct drr_spill {
294			uint64_t drr_object;
295			uint64_t drr_length;
296			uint64_t drr_toguid;
297			uint8_t drr_flags;
298			uint8_t drr_compressiontype;
299			uint8_t drr_pad[6];
300			/* only nonzero for raw streams */
301			uint64_t drr_compressed_size;
302			uint8_t drr_salt[ZIO_DATA_SALT_LEN];
303			uint8_t drr_iv[ZIO_DATA_IV_LEN];
304			uint8_t drr_mac[ZIO_DATA_MAC_LEN];
305			dmu_object_type_t drr_type;
306			/* spill data follows */
307		} drr_spill;
308		struct drr_write_embedded {
309			uint64_t drr_object;
310			uint64_t drr_offset;
311			/* logical length, should equal blocksize */
312			uint64_t drr_length;
313			uint64_t drr_toguid;
314			uint8_t drr_compression;
315			uint8_t drr_etype;
316			uint8_t drr_pad[6];
317			uint32_t drr_lsize; /* uncompressed size of payload */
318			uint32_t drr_psize; /* compr. (real) size of payload */
319			/* (possibly compressed) content follows */
320		} drr_write_embedded;
321		struct drr_object_range {
322			uint64_t drr_firstobj;
323			uint64_t drr_numslots;
324			uint64_t drr_toguid;
325			uint8_t drr_salt[ZIO_DATA_SALT_LEN];
326			uint8_t drr_iv[ZIO_DATA_IV_LEN];
327			uint8_t drr_mac[ZIO_DATA_MAC_LEN];
328			uint8_t drr_flags;
329			uint8_t drr_pad[3];
330		} drr_object_range;
331
332		/*
333		 * Nore: drr_checksum is overlaid with all record types
334		 * except DRR_BEGIN.  Therefore its (non-pad) members
335		 * must not overlap with members from the other structs.
336		 * We accomplish this by putting its members at the very
337		 * end of the struct.
338		 */
339		struct drr_checksum {
340			uint64_t drr_pad[34];
341			/*
342			 * fletcher-4 checksum of everything preceding the
343			 * checksum.
344			 */
345			zio_cksum_t drr_checksum;
346		} drr_checksum;
347	} drr_u;
348} dmu_replay_record_t;
349
350/* diff record range types */
351typedef enum diff_type {
352	DDR_NONE = 0x1,
353	DDR_INUSE = 0x2,
354	DDR_FREE = 0x4
355} diff_type_t;
356
357/*
358 * The diff reports back ranges of free or in-use objects.
359 */
360typedef struct dmu_diff_record {
361	uint64_t ddr_type;
362	uint64_t ddr_first;
363	uint64_t ddr_last;
364} dmu_diff_record_t;
365
366typedef struct zinject_record {
367	uint64_t	zi_objset;
368	uint64_t	zi_object;
369	uint64_t	zi_start;
370	uint64_t	zi_end;
371	uint64_t	zi_guid;
372	uint32_t	zi_level;
373	uint32_t	zi_error;
374	uint64_t	zi_type;
375	uint32_t	zi_freq;
376	uint32_t	zi_failfast;
377	char		zi_func[MAXNAMELEN];
378	uint32_t	zi_iotype;
379	int32_t		zi_duration;
380	uint64_t	zi_timer;
381	uint64_t	zi_nlanes;
382	uint32_t	zi_cmd;
383	uint32_t	zi_dvas;
384} zinject_record_t;
385
386#define	ZINJECT_NULL		0x1
387#define	ZINJECT_FLUSH_ARC	0x2
388#define	ZINJECT_UNLOAD_SPA	0x4
389#define	ZINJECT_CALC_RANGE	0x8
390
391#define	ZI_NO_DVA		(-1)
392
393/* scaled frequency ranges */
394#define	ZI_PERCENTAGE_MIN	4294UL
395#define	ZI_PERCENTAGE_MAX	UINT32_MAX
396
397typedef enum zinject_type {
398	ZINJECT_UNINITIALIZED,
399	ZINJECT_DATA_FAULT,
400	ZINJECT_DEVICE_FAULT,
401	ZINJECT_LABEL_FAULT,
402	ZINJECT_IGNORED_WRITES,
403	ZINJECT_PANIC,
404	ZINJECT_DELAY_IO,
405	ZINJECT_DECRYPT_FAULT,
406} zinject_type_t;
407
408typedef struct zfs_share {
409	uint64_t	z_exportdata;
410	uint64_t	z_sharedata;
411	uint64_t	z_sharetype;	/* 0 = share, 1 = unshare */
412	uint64_t	z_sharemax;  /* max length of share string */
413} zfs_share_t;
414
415/*
416 * ZFS file systems may behave the usual, POSIX-compliant way, where
417 * name lookups are case-sensitive.  They may also be set up so that
418 * all the name lookups are case-insensitive, or so that only some
419 * lookups, the ones that set an FIGNORECASE flag, are case-insensitive.
420 */
421typedef enum zfs_case {
422	ZFS_CASE_SENSITIVE,
423	ZFS_CASE_INSENSITIVE,
424	ZFS_CASE_MIXED
425} zfs_case_t;
426
427/*
428 * Note: this struct must have the same layout in 32-bit and 64-bit, so
429 * that 32-bit processes (like /sbin/zfs) can pass it to the 64-bit
430 * kernel.  Therefore, we add padding to it so that no "hidden" padding
431 * is automatically added on 64-bit (but not on 32-bit).
432 */
433typedef struct zfs_cmd {
434	char		zc_name[MAXPATHLEN];	/* name of pool or dataset */
435	uint64_t	zc_nvlist_src;		/* really (char *) */
436	uint64_t	zc_nvlist_src_size;
437	uint64_t	zc_nvlist_dst;		/* really (char *) */
438	uint64_t	zc_nvlist_dst_size;
439	boolean_t	zc_nvlist_dst_filled;	/* put an nvlist in dst? */
440	int		zc_pad2;
441
442	/*
443	 * The following members are for legacy ioctls which haven't been
444	 * converted to the new method.
445	 */
446	uint64_t	zc_history;		/* really (char *) */
447	char		zc_value[MAXPATHLEN * 2];
448	char		zc_string[MAXNAMELEN];
449	uint64_t	zc_guid;
450	uint64_t	zc_nvlist_conf;		/* really (char *) */
451	uint64_t	zc_nvlist_conf_size;
452	uint64_t	zc_cookie;
453	uint64_t	zc_objset_type;
454	uint64_t	zc_perm_action;
455	uint64_t	zc_history_len;
456	uint64_t	zc_history_offset;
457	uint64_t	zc_obj;
458	uint64_t	zc_iflags;		/* internal to zfs(7fs) */
459	zfs_share_t	zc_share;
460	dmu_objset_stats_t zc_objset_stats;
461	dmu_replay_record_t zc_begin_record;
462	zinject_record_t zc_inject_record;
463	uint32_t	zc_defer_destroy;
464	uint32_t	zc_flags;
465	uint64_t	zc_action_handle;
466	int		zc_cleanup_fd;
467	uint8_t		zc_simple;
468	uint8_t		zc_pad3[3];
469	boolean_t	zc_resumable;
470	uint32_t	zc_pad4;
471	uint64_t	zc_sendobj;
472	uint64_t	zc_fromobj;
473	uint64_t	zc_createtxg;
474	zfs_stat_t	zc_stat;
475} zfs_cmd_t;
476
477typedef struct zfs_useracct {
478	char zu_domain[256];
479	uid_t zu_rid;
480	uint32_t zu_pad;
481	uint64_t zu_space;
482} zfs_useracct_t;
483
484#define	ZFSDEV_MAX_MINOR	(1 << 16)
485#define	ZFS_MIN_MINOR	(ZFSDEV_MAX_MINOR + 1)
486
487#define	ZPOOL_EXPORT_AFTER_SPLIT 0x1
488
489#ifdef _KERNEL
490struct objset;
491struct zfsvfs;
492
493typedef struct zfs_creat {
494	nvlist_t	*zct_zplprops;
495	nvlist_t	*zct_props;
496} zfs_creat_t;
497
498extern dev_info_t *zfs_dip;
499
500extern int zfs_secpolicy_snapshot_perms(const char *, cred_t *);
501extern int zfs_secpolicy_rename_perms(const char *, const char *, cred_t *);
502extern int zfs_secpolicy_destroy_perms(const char *, cred_t *);
503extern int zfs_busy(void);
504extern void zfs_unmount_snap(const char *);
505extern void zfs_destroy_unmount_origin(const char *);
506extern int getzfsvfs_impl(struct objset *, struct zfsvfs **);
507extern int getzfsvfs(const char *, struct zfsvfs **);
508
509/*
510 * ZFS minor numbers can refer to either a control device instance or
511 * a zvol. Depending on the value of zss_type, zss_data points to either
512 * a zvol_state_t or a zfs_onexit_t.
513 */
514enum zfs_soft_state_type {
515	ZSST_ZVOL,
516	ZSST_CTLDEV
517};
518
519typedef struct zfs_soft_state {
520	enum zfs_soft_state_type zss_type;
521	void *zss_data;
522} zfs_soft_state_t;
523
524extern void *zfsdev_get_soft_state(minor_t minor,
525    enum zfs_soft_state_type which);
526extern minor_t zfsdev_minor_alloc(void);
527
528extern void *zfsdev_state;
529extern kmutex_t zfsdev_state_lock;
530
531#endif	/* _KERNEL */
532
533#ifdef	__cplusplus
534}
535#endif
536
537#endif	/* _SYS_ZFS_IOCTL_H */
538