zfs_ioctl.h revision 307053 
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, 2015 by Delphix. All rights reserved.
25 * Copyright 2016 RackTop Systems.
26 * Copyright (c) 2014 Integros [integros.com]
27 */
28
29#ifndef	_SYS_ZFS_IOCTL_H
30#define	_SYS_ZFS_IOCTL_H
31
32#include <sys/cred.h>
33#include <sys/dmu.h>
34#include <sys/zio.h>
35#include <sys/dsl_deleg.h>
36#include <sys/spa.h>
37#include <sys/zfs_stat.h>
38
39#ifdef _KERNEL
40#include <sys/nvpair.h>
41#endif	/* _KERNEL */
42
43#ifdef	__cplusplus
44extern "C" {
45#endif
46
47/*
48 * The structures in this file are passed between userland and the
49 * kernel.  Userland may be running a 32-bit process, while the kernel
50 * is 64-bit.  Therefore, these structures need to compile the same in
51 * 32-bit and 64-bit.  This means not using type "long", and adding
52 * explicit padding so that the 32-bit structure will not be packed more
53 * tightly than the 64-bit structure (which requires 64-bit alignment).
54 */
55
56/*
57 * Property values for snapdir
58 */
59#define	ZFS_SNAPDIR_HIDDEN		0
60#define	ZFS_SNAPDIR_VISIBLE		1
61
62/*
63 * Field manipulation macros for the drr_versioninfo field of the
64 * send stream header.
65 */
66
67/*
68 * Header types for zfs send streams.
69 */
70typedef enum drr_headertype {
71	DMU_SUBSTREAM = 0x1,
72	DMU_COMPOUNDSTREAM = 0x2
73} drr_headertype_t;
74
75#define	DMU_GET_STREAM_HDRTYPE(vi)	BF64_GET((vi), 0, 2)
76#define	DMU_SET_STREAM_HDRTYPE(vi, x)	BF64_SET((vi), 0, 2, x)
77
78#define	DMU_GET_FEATUREFLAGS(vi)	BF64_GET((vi), 2, 30)
79#define	DMU_SET_FEATUREFLAGS(vi, x)	BF64_SET((vi), 2, 30, x)
80
81/*
82 * Feature flags for zfs send streams (flags in drr_versioninfo)
83 */
84
85#define	DMU_BACKUP_FEATURE_DEDUP		(1 << 0)
86#define	DMU_BACKUP_FEATURE_DEDUPPROPS		(1 << 1)
87#define	DMU_BACKUP_FEATURE_SA_SPILL		(1 << 2)
88/* flags #3 - #15 are reserved for incompatible closed-source implementations */
89#define	DMU_BACKUP_FEATURE_EMBED_DATA		(1 << 16)
90#define	DMU_BACKUP_FEATURE_EMBED_DATA_LZ4	(1 << 17)
91/* flag #18 is reserved for a Delphix feature */
92#define	DMU_BACKUP_FEATURE_LARGE_BLOCKS		(1 << 19)
93#define	DMU_BACKUP_FEATURE_RESUMING		(1 << 20)
94
95/*
96 * Mask of all supported backup features
97 */
98#define	DMU_BACKUP_FEATURE_MASK	(DMU_BACKUP_FEATURE_DEDUP | \
99    DMU_BACKUP_FEATURE_DEDUPPROPS | DMU_BACKUP_FEATURE_SA_SPILL | \
100    DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_EMBED_DATA_LZ4 | \
101    DMU_BACKUP_FEATURE_RESUMING | \
102    DMU_BACKUP_FEATURE_LARGE_BLOCKS)
103
104/* Are all features in the given flag word currently supported? */
105#define	DMU_STREAM_SUPPORTED(x)	(!((x) & ~DMU_BACKUP_FEATURE_MASK))
106
107typedef enum dmu_send_resume_token_version {
108	ZFS_SEND_RESUME_TOKEN_VERSION = 1
109} dmu_send_resume_token_version_t;
110
111/*
112 * The drr_versioninfo field of the dmu_replay_record has the
113 * following layout:
114 *
115 *	64	56	48	40	32	24	16	8	0
116 *	+-------+-------+-------+-------+-------+-------+-------+-------+
117 *  	|		reserved	|        feature-flags	    |C|S|
118 *	+-------+-------+-------+-------+-------+-------+-------+-------+
119 *
120 * The low order two bits indicate the header type: SUBSTREAM (0x1)
121 * or COMPOUNDSTREAM (0x2).  Using two bits for this is historical:
122 * this field used to be a version number, where the two version types
123 * were 1 and 2.  Using two bits for this allows earlier versions of
124 * the code to be able to recognize send streams that don't use any
125 * of the features indicated by feature flags.
126 */
127
128#define	DMU_BACKUP_MAGIC 0x2F5bacbacULL
129
130/*
131 * Send stream flags.  Bits 24-31 are reserved for vendor-specific
132 * implementations and should not be used.
133 */
134#define	DRR_FLAG_CLONE		(1<<0)
135#define	DRR_FLAG_CI_DATA	(1<<1)
136/*
137 * This send stream, if it is a full send, includes the FREE and FREEOBJECT
138 * records that are created by the sending process.  This means that the send
139 * stream can be received as a clone, even though it is not an incremental.
140 * This is not implemented as a feature flag, because the receiving side does
141 * not need to have implemented it to receive this stream; it is fully backwards
142 * compatible.  We need a flag, though, because full send streams without it
143 * cannot necessarily be received as a clone correctly.
144 */
145#define	DRR_FLAG_FREERECORDS	(1<<2)
146
147/*
148 * flags in the drr_checksumflags field in the DRR_WRITE and
149 * DRR_WRITE_BYREF blocks
150 */
151#define	DRR_CHECKSUM_DEDUP	(1<<0)
152
153#define	DRR_IS_DEDUP_CAPABLE(flags)	((flags) & DRR_CHECKSUM_DEDUP)
154
155/*
156 * zfs ioctl command structure
157 */
158struct drr_begin {
159	uint64_t drr_magic;
160	uint64_t drr_versioninfo; /* was drr_version */
161	uint64_t drr_creation_time;
162	dmu_objset_type_t drr_type;
163	uint32_t drr_flags;
164	uint64_t drr_toguid;
165	uint64_t drr_fromguid;
166	char drr_toname[MAXNAMELEN];
167};
168
169struct drr_end {
170	zio_cksum_t drr_checksum;
171	uint64_t drr_toguid;
172};
173
174struct drr_object {
175	uint64_t drr_object;
176	dmu_object_type_t drr_type;
177	dmu_object_type_t drr_bonustype;
178	uint32_t drr_blksz;
179	uint32_t drr_bonuslen;
180	uint8_t drr_checksumtype;
181	uint8_t drr_compress;
182	uint8_t drr_pad[6];
183	uint64_t drr_toguid;
184	/* bonus content follows */
185};
186
187struct drr_freeobjects {
188	uint64_t drr_firstobj;
189	uint64_t drr_numobjs;
190	uint64_t drr_toguid;
191};
192
193struct drr_write {
194	uint64_t drr_object;
195	dmu_object_type_t drr_type;
196	uint32_t drr_pad;
197	uint64_t drr_offset;
198	uint64_t drr_length;
199	uint64_t drr_toguid;
200	uint8_t drr_checksumtype;
201	uint8_t drr_checksumflags;
202	uint8_t drr_pad2[6];
203	ddt_key_t drr_key; /* deduplication key */
204	/* content follows */
205};
206
207struct drr_free {
208	uint64_t drr_object;
209	uint64_t drr_offset;
210	uint64_t drr_length;
211	uint64_t drr_toguid;
212};
213
214struct drr_write_byref {
215	/* where to put the data */
216	uint64_t drr_object;
217	uint64_t drr_offset;
218	uint64_t drr_length;
219	uint64_t drr_toguid;
220	/* where to find the prior copy of the data */
221	uint64_t drr_refguid;
222	uint64_t drr_refobject;
223	uint64_t drr_refoffset;
224	/* properties of the data */
225	uint8_t drr_checksumtype;
226	uint8_t drr_checksumflags;
227	uint8_t drr_pad2[6];
228	ddt_key_t drr_key; /* deduplication key */
229};
230
231struct drr_spill {
232	uint64_t drr_object;
233	uint64_t drr_length;
234	uint64_t drr_toguid;
235	uint64_t drr_pad[4]; /* needed for crypto */
236	/* spill data follows */
237};
238
239typedef struct dmu_replay_record {
240	enum {
241		DRR_BEGIN, DRR_OBJECT, DRR_FREEOBJECTS,
242		DRR_WRITE, DRR_FREE, DRR_END, DRR_WRITE_BYREF,
243		DRR_SPILL, DRR_WRITE_EMBEDDED, DRR_NUMTYPES
244	} drr_type;
245	uint32_t drr_payloadlen;
246	union {
247		struct drr_begin drr_begin;
248		struct drr_end drr_end;
249		struct drr_object drr_object;
250		struct drr_freeobjects drr_freeobjects;
251		struct drr_write drr_write;
252		struct drr_free drr_free;
253		struct drr_write_byref drr_write_byref;
254		struct drr_spill drr_spill;
255		struct drr_write_embedded {
256			uint64_t drr_object;
257			uint64_t drr_offset;
258			/* logical length, should equal blocksize */
259			uint64_t drr_length;
260			uint64_t drr_toguid;
261			uint8_t drr_compression;
262			uint8_t drr_etype;
263			uint8_t drr_pad[6];
264			uint32_t drr_lsize; /* uncompressed size of payload */
265			uint32_t drr_psize; /* compr. (real) size of payload */
266			/* (possibly compressed) content follows */
267		} drr_write_embedded;
268
269		/*
270		 * Nore: drr_checksum is overlaid with all record types
271		 * except DRR_BEGIN.  Therefore its (non-pad) members
272		 * must not overlap with members from the other structs.
273		 * We accomplish this by putting its members at the very
274		 * end of the struct.
275		 */
276		struct drr_checksum {
277			uint64_t drr_pad[34];
278			/*
279			 * fletcher-4 checksum of everything preceding the
280			 * checksum.
281			 */
282			zio_cksum_t drr_checksum;
283		} drr_checksum;
284	} drr_u;
285} dmu_replay_record_t;
286
287/* diff record range types */
288typedef enum diff_type {
289	DDR_NONE = 0x1,
290	DDR_INUSE = 0x2,
291	DDR_FREE = 0x4
292} diff_type_t;
293
294/*
295 * The diff reports back ranges of free or in-use objects.
296 */
297typedef struct dmu_diff_record {
298	uint64_t ddr_type;
299	uint64_t ddr_first;
300	uint64_t ddr_last;
301} dmu_diff_record_t;
302
303typedef struct zinject_record {
304	uint64_t	zi_objset;
305	uint64_t	zi_object;
306	uint64_t	zi_start;
307	uint64_t	zi_end;
308	uint64_t	zi_guid;
309	uint32_t	zi_level;
310	uint32_t	zi_error;
311	uint64_t	zi_type;
312	uint32_t	zi_freq;
313	uint32_t	zi_failfast;
314	char		zi_func[MAXNAMELEN];
315	uint32_t	zi_iotype;
316	int32_t		zi_duration;
317	uint64_t	zi_timer;
318	uint64_t	zi_nlanes;
319	uint32_t	zi_cmd;
320	uint32_t	zi_pad;
321} zinject_record_t;
322
323#define	ZINJECT_NULL		0x1
324#define	ZINJECT_FLUSH_ARC	0x2
325#define	ZINJECT_UNLOAD_SPA	0x4
326
327typedef enum zinject_type {
328	ZINJECT_UNINITIALIZED,
329	ZINJECT_DATA_FAULT,
330	ZINJECT_DEVICE_FAULT,
331	ZINJECT_LABEL_FAULT,
332	ZINJECT_IGNORED_WRITES,
333	ZINJECT_PANIC,
334	ZINJECT_DELAY_IO,
335} zinject_type_t;
336
337typedef struct zfs_share {
338	uint64_t	z_exportdata;
339	uint64_t	z_sharedata;
340	uint64_t	z_sharetype;	/* 0 = share, 1 = unshare */
341	uint64_t	z_sharemax;  /* max length of share string */
342} zfs_share_t;
343
344/*
345 * ZFS file systems may behave the usual, POSIX-compliant way, where
346 * name lookups are case-sensitive.  They may also be set up so that
347 * all the name lookups are case-insensitive, or so that only some
348 * lookups, the ones that set an FIGNORECASE flag, are case-insensitive.
349 */
350typedef enum zfs_case {
351	ZFS_CASE_SENSITIVE,
352	ZFS_CASE_INSENSITIVE,
353	ZFS_CASE_MIXED
354} zfs_case_t;
355
356typedef struct zfs_cmd {
357	char		zc_name[MAXPATHLEN];	/* name of pool or dataset */
358	uint64_t	zc_nvlist_src;		/* really (char *) */
359	uint64_t	zc_nvlist_src_size;
360	uint64_t	zc_nvlist_dst;		/* really (char *) */
361	uint64_t	zc_nvlist_dst_size;
362	boolean_t	zc_nvlist_dst_filled;	/* put an nvlist in dst? */
363	int		zc_pad2;
364
365	/*
366	 * The following members are for legacy ioctls which haven't been
367	 * converted to the new method.
368	 */
369	uint64_t	zc_history;		/* really (char *) */
370	char		zc_value[MAXPATHLEN * 2];
371	char		zc_string[MAXNAMELEN];
372	uint64_t	zc_guid;
373	uint64_t	zc_nvlist_conf;		/* really (char *) */
374	uint64_t	zc_nvlist_conf_size;
375	uint64_t	zc_cookie;
376	uint64_t	zc_objset_type;
377	uint64_t	zc_perm_action;
378	uint64_t	zc_history_len;
379	uint64_t	zc_history_offset;
380	uint64_t	zc_obj;
381	uint64_t	zc_iflags;		/* internal to zfs(7fs) */
382	zfs_share_t	zc_share;
383	uint64_t	zc_jailid;
384	dmu_objset_stats_t zc_objset_stats;
385	dmu_replay_record_t zc_begin_record;
386	zinject_record_t zc_inject_record;
387	uint32_t	zc_defer_destroy;
388	uint32_t	zc_flags;
389	uint64_t	zc_action_handle;
390	int		zc_cleanup_fd;
391	uint8_t		zc_simple;
392	boolean_t	zc_resumable;
393	uint64_t	zc_sendobj;
394	uint64_t	zc_fromobj;
395	uint64_t	zc_createtxg;
396	zfs_stat_t	zc_stat;
397} zfs_cmd_t;
398
399typedef struct zfs_useracct {
400	char zu_domain[256];
401	uid_t zu_rid;
402	uint32_t zu_pad;
403	uint64_t zu_space;
404} zfs_useracct_t;
405
406#define	ZFSDEV_MAX_MINOR	(1 << 16)
407#define	ZFS_MIN_MINOR	(ZFSDEV_MAX_MINOR + 1)
408
409#define	ZPOOL_EXPORT_AFTER_SPLIT 0x1
410
411#ifdef _KERNEL
412
413typedef struct zfs_creat {
414	nvlist_t	*zct_zplprops;
415	nvlist_t	*zct_props;
416} zfs_creat_t;
417
418extern int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr);
419extern int zfs_secpolicy_rename_perms(const char *from,
420    const char *to, cred_t *cr);
421extern int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr);
422extern int zfs_busy(void);
423extern int zfs_unmount_snap(const char *);
424extern void zfs_destroy_unmount_origin(const char *);
425
426/*
427 * ZFS minor numbers can refer to either a control device instance or
428 * a zvol. Depending on the value of zss_type, zss_data points to either
429 * a zvol_state_t or a zfs_onexit_t.
430 */
431enum zfs_soft_state_type {
432	ZSST_ZVOL,
433	ZSST_CTLDEV
434};
435
436typedef struct zfs_soft_state {
437	enum zfs_soft_state_type zss_type;
438	void *zss_data;
439} zfs_soft_state_t;
440
441extern void *zfsdev_get_soft_state(minor_t minor,
442    enum zfs_soft_state_type which);
443extern minor_t zfsdev_minor_alloc(void);
444
445extern void *zfsdev_state;
446
447#endif	/* _KERNEL */
448
449#ifdef	__cplusplus
450}
451#endif
452
453#endif	/* _SYS_ZFS_IOCTL_H */
454