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 Nexenta Systems, Inc. All rights reserved.
  26  * Copyright 2016 RackTop Systems.
  27  * Copyright (c) 2014 Integros [integros.com]
  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
  45 extern "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  */
  71 typedef 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 a Delphix feature */
  96 #define DMU_BACKUP_FEATURE_COMPRESSED           (1 << 22)
  97 /* flag #23 is reserved for the large dnode feature */
  98 
  99 /*
 100  * Mask of all supported backup features
 101  */
 102 #define DMU_BACKUP_FEATURE_MASK (DMU_BACKUP_FEATURE_DEDUP | \
 103     DMU_BACKUP_FEATURE_DEDUPPROPS | DMU_BACKUP_FEATURE_SA_SPILL | \
 104     DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_LZ4 | \
 105     DMU_BACKUP_FEATURE_RESUMING | \
 106     DMU_BACKUP_FEATURE_LARGE_BLOCKS | \
 107     DMU_BACKUP_FEATURE_COMPRESSED)
 108 
 109 /* Are all features in the given flag word currently supported? */
 110 #define DMU_STREAM_SUPPORTED(x) (!((x) & ~DMU_BACKUP_FEATURE_MASK))
 111 
 112 typedef enum dmu_send_resume_token_version {
 113         ZFS_SEND_RESUME_TOKEN_VERSION = 1
 114 } dmu_send_resume_token_version_t;
 115 
 116 /*
 117  * The drr_versioninfo field of the dmu_replay_record has the
 118  * following layout:
 119  *
 120  *      64      56      48      40      32      24      16      8       0
 121  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 122  *      |               reserved        |        feature-flags      |C|S|
 123  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 124  *
 125  * The low order two bits indicate the header type: SUBSTREAM (0x1)
 126  * or COMPOUNDSTREAM (0x2).  Using two bits for this is historical:
 127  * this field used to be a version number, where the two version types
 128  * were 1 and 2.  Using two bits for this allows earlier versions of
 129  * the code to be able to recognize send streams that don't use any
 130  * of the features indicated by feature flags.
 131  */
 132 
 133 #define DMU_BACKUP_MAGIC 0x2F5bacbacULL
 134 
 135 /*
 136  * Send stream flags.  Bits 24-31 are reserved for vendor-specific
 137  * implementations and should not be used.
 138  */
 139 #define DRR_FLAG_CLONE          (1<<0)
 140 #define DRR_FLAG_CI_DATA        (1<<1)
 141 /*
 142  * This send stream, if it is a full send, includes the FREE and FREEOBJECT
 143  * records that are created by the sending process.  This means that the send
 144  * stream can be received as a clone, even though it is not an incremental.
 145  * This is not implemented as a feature flag, because the receiving side does
 146  * not need to have implemented it to receive this stream; it is fully backwards
 147  * compatible.  We need a flag, though, because full send streams without it
 148  * cannot necessarily be received as a clone correctly.
 149  */
 150 #define DRR_FLAG_FREERECORDS    (1<<2)
 151 
 152 /*
 153  * flags in the drr_checksumflags field in the DRR_WRITE and
 154  * DRR_WRITE_BYREF blocks
 155  */
 156 #define DRR_CHECKSUM_DEDUP      (1<<0)
 157 
 158 #define DRR_IS_DEDUP_CAPABLE(flags)     ((flags) & DRR_CHECKSUM_DEDUP)
 159 
 160 /* deal with compressed drr_write replay records */
 161 #define DRR_WRITE_COMPRESSED(drrw)      ((drrw)->drr_compressiontype != 0)
 162 #define DRR_WRITE_PAYLOAD_SIZE(drrw) \
 163         (DRR_WRITE_COMPRESSED(drrw) ? (drrw)->drr_compressed_size : \
 164         (drrw)->drr_logical_size)
 165 
 166 /*
 167  * zfs ioctl command structure
 168  */
 169 typedef struct dmu_replay_record {
 170         enum {
 171                 DRR_BEGIN, DRR_OBJECT, DRR_FREEOBJECTS,
 172                 DRR_WRITE, DRR_FREE, DRR_END, DRR_WRITE_BYREF,
 173                 DRR_SPILL, DRR_WRITE_EMBEDDED, DRR_NUMTYPES
 174         } drr_type;
 175         uint32_t drr_payloadlen;
 176         union {
 177                 struct drr_begin {
 178                         uint64_t drr_magic;
 179                         uint64_t drr_versioninfo; /* was drr_version */
 180                         uint64_t drr_creation_time;
 181                         dmu_objset_type_t drr_type;
 182                         uint32_t drr_flags;
 183                         uint64_t drr_toguid;
 184                         uint64_t drr_fromguid;
 185                         char drr_toname[MAXNAMELEN];
 186                 } drr_begin;
 187                 struct drr_end {
 188                         zio_cksum_t drr_checksum;
 189                         uint64_t drr_toguid;
 190                 } drr_end;
 191                 struct drr_object {
 192                         uint64_t drr_object;
 193                         dmu_object_type_t drr_type;
 194                         dmu_object_type_t drr_bonustype;
 195                         uint32_t drr_blksz;
 196                         uint32_t drr_bonuslen;
 197                         uint8_t drr_checksumtype;
 198                         uint8_t drr_compress;
 199                         uint8_t drr_pad[6];
 200                         uint64_t drr_toguid;
 201                         /* bonus content follows */
 202                 } drr_object;
 203                 struct drr_freeobjects {
 204                         uint64_t drr_firstobj;
 205                         uint64_t drr_numobjs;
 206                         uint64_t drr_toguid;
 207                 } drr_freeobjects;
 208                 struct drr_write {
 209                         uint64_t drr_object;
 210                         dmu_object_type_t drr_type;
 211                         uint32_t drr_pad;
 212                         uint64_t drr_offset;
 213                         uint64_t drr_logical_size;
 214                         uint64_t drr_toguid;
 215                         uint8_t drr_checksumtype;
 216                         uint8_t drr_checksumflags;
 217                         uint8_t drr_compressiontype;
 218                         uint8_t drr_pad2[5];
 219                         /* deduplication key */
 220                         ddt_key_t drr_key;
 221                         /* only nonzero if drr_compressiontype is not 0 */
 222                         uint64_t drr_compressed_size;
 223                         /* content follows */
 224                 } drr_write;
 225                 struct drr_free {
 226                         uint64_t drr_object;
 227                         uint64_t drr_offset;
 228                         uint64_t drr_length;
 229                         uint64_t drr_toguid;
 230                 } drr_free;
 231                 struct drr_write_byref {
 232                         /* where to put the data */
 233                         uint64_t drr_object;
 234                         uint64_t drr_offset;
 235                         uint64_t drr_length;
 236                         uint64_t drr_toguid;
 237                         /* where to find the prior copy of the data */
 238                         uint64_t drr_refguid;
 239                         uint64_t drr_refobject;
 240                         uint64_t drr_refoffset;
 241                         /* properties of the data */
 242                         uint8_t drr_checksumtype;
 243                         uint8_t drr_checksumflags;
 244                         uint8_t drr_pad2[6];
 245                         ddt_key_t drr_key; /* deduplication key */
 246                 } drr_write_byref;
 247                 struct drr_spill {
 248                         uint64_t drr_object;
 249                         uint64_t drr_length;
 250                         uint64_t drr_toguid;
 251                         uint64_t drr_pad[4]; /* needed for crypto */
 252                         /* spill data follows */
 253                 } drr_spill;
 254                 struct drr_write_embedded {
 255                         uint64_t drr_object;
 256                         uint64_t drr_offset;
 257                         /* logical length, should equal blocksize */
 258                         uint64_t drr_length;
 259                         uint64_t drr_toguid;
 260                         uint8_t drr_compression;
 261                         uint8_t drr_etype;
 262                         uint8_t drr_pad[6];
 263                         uint32_t drr_lsize; /* uncompressed size of payload */
 264                         uint32_t drr_psize; /* compr. (real) size of payload */
 265                         /* (possibly compressed) content follows */
 266                 } drr_write_embedded;
 267 
 268                 /*
 269                  * Nore: drr_checksum is overlaid with all record types
 270                  * except DRR_BEGIN.  Therefore its (non-pad) members
 271                  * must not overlap with members from the other structs.
 272                  * We accomplish this by putting its members at the very
 273                  * end of the struct.
 274                  */
 275                 struct drr_checksum {
 276                         uint64_t drr_pad[34];
 277                         /*
 278                          * fletcher-4 checksum of everything preceding the
 279                          * checksum.
 280                          */
 281                         zio_cksum_t drr_checksum;
 282                 } drr_checksum;
 283         } drr_u;
 284 } dmu_replay_record_t;
 285 
 286 /* diff record range types */
 287 typedef enum diff_type {
 288         DDR_NONE = 0x1,
 289         DDR_INUSE = 0x2,
 290         DDR_FREE = 0x4
 291 } diff_type_t;
 292 
 293 /*
 294  * The diff reports back ranges of free or in-use objects.
 295  */
 296 typedef struct dmu_diff_record {
 297         uint64_t ddr_type;
 298         uint64_t ddr_first;
 299         uint64_t ddr_last;
 300 } dmu_diff_record_t;
 301 
 302 typedef struct zinject_record {
 303         uint64_t        zi_objset;
 304         uint64_t        zi_object;
 305         uint64_t        zi_start;
 306         uint64_t        zi_end;
 307         uint64_t        zi_guid;
 308         uint32_t        zi_level;
 309         uint32_t        zi_error;
 310         uint64_t        zi_type;
 311         uint32_t        zi_freq;
 312         uint32_t        zi_failfast;
 313         char            zi_func[MAXNAMELEN];
 314         uint32_t        zi_iotype;
 315         int32_t         zi_duration;
 316         uint64_t        zi_timer;
 317         uint64_t        zi_nlanes;
 318         uint32_t        zi_cmd;
 319         uint32_t        zi_pad;
 320 } zinject_record_t;
 321 
 322 #define ZINJECT_NULL            0x1
 323 #define ZINJECT_FLUSH_ARC       0x2
 324 #define ZINJECT_UNLOAD_SPA      0x4
 325 
 326 typedef enum zinject_type {
 327         ZINJECT_UNINITIALIZED,
 328         ZINJECT_DATA_FAULT,
 329         ZINJECT_DEVICE_FAULT,
 330         ZINJECT_LABEL_FAULT,
 331         ZINJECT_IGNORED_WRITES,
 332         ZINJECT_PANIC,
 333         ZINJECT_DELAY_IO,
 334 } zinject_type_t;
 335 
 336 typedef struct zfs_share {
 337         uint64_t        z_exportdata;
 338         uint64_t        z_sharedata;
 339         uint64_t        z_sharetype;    /* 0 = share, 1 = unshare */
 340         uint64_t        z_sharemax;  /* max length of share string */
 341 } zfs_share_t;
 342 
 343 /*
 344  * ZFS file systems may behave the usual, POSIX-compliant way, where
 345  * name lookups are case-sensitive.  They may also be set up so that
 346  * all the name lookups are case-insensitive, or so that only some
 347  * lookups, the ones that set an FIGNORECASE flag, are case-insensitive.
 348  */
 349 typedef enum zfs_case {
 350         ZFS_CASE_SENSITIVE,
 351         ZFS_CASE_INSENSITIVE,
 352         ZFS_CASE_MIXED
 353 } zfs_case_t;
 354 
 355 /*
 356  * Note: this struct must have the same layout in 32-bit and 64-bit, so
 357  * that 32-bit processes (like /sbin/zfs) can pass it to the 64-bit
 358  * kernel.  Therefore, we add padding to it so that no "hidden" padding
 359  * is automatically added on 64-bit (but not on 32-bit).
 360  */
 361 typedef struct zfs_cmd {
 362         char            zc_name[MAXPATHLEN];    /* name of pool or dataset */
 363         uint64_t        zc_nvlist_src;          /* really (char *) */
 364         uint64_t        zc_nvlist_src_size;
 365         uint64_t        zc_nvlist_dst;          /* really (char *) */
 366         uint64_t        zc_nvlist_dst_size;
 367         boolean_t       zc_nvlist_dst_filled;   /* put an nvlist in dst? */
 368         int             zc_pad2;
 369 
 370         /*
 371          * The following members are for legacy ioctls which haven't been
 372          * converted to the new method.
 373          */
 374         uint64_t        zc_history;             /* really (char *) */
 375         char            zc_value[MAXPATHLEN * 2];
 376         char            zc_string[MAXNAMELEN];
 377         uint64_t        zc_guid;
 378         uint64_t        zc_nvlist_conf;         /* really (char *) */
 379         uint64_t        zc_nvlist_conf_size;
 380         uint64_t        zc_cookie;
 381         uint64_t        zc_objset_type;
 382         uint64_t        zc_perm_action;
 383         uint64_t        zc_history_len;
 384         uint64_t        zc_history_offset;
 385         uint64_t        zc_obj;
 386         uint64_t        zc_iflags;              /* internal to zfs(7fs) */
 387         zfs_share_t     zc_share;
 388         dmu_objset_stats_t zc_objset_stats;
 389         dmu_replay_record_t zc_begin_record;
 390         zinject_record_t zc_inject_record;
 391         uint32_t        zc_defer_destroy;
 392         uint32_t        zc_flags;
 393         uint64_t        zc_action_handle;
 394         int             zc_cleanup_fd;
 395         uint8_t         zc_simple;
 396         uint8_t         zc_pad3[3];
 397         boolean_t       zc_resumable;
 398         uint32_t        zc_pad4;
 399         uint64_t        zc_sendobj;
 400         uint64_t        zc_fromobj;
 401         uint64_t        zc_createtxg;
 402         zfs_stat_t      zc_stat;
 403         uint64_t        zc_sendcounter;
 404         boolean_t       zc_sendsize;
 405 } zfs_cmd_t;
 406 
 407 typedef struct zfs_useracct {
 408         char zu_domain[256];
 409         uid_t zu_rid;
 410         uint32_t zu_pad;
 411         uint64_t zu_space;
 412 } zfs_useracct_t;
 413 
 414 #define ZFSDEV_MAX_MINOR        (1 << 16)
 415 #define ZFS_MIN_MINOR   (ZFSDEV_MAX_MINOR + 1)
 416 
 417 #define ZPOOL_EXPORT_AFTER_SPLIT 0x1
 418 
 419 #define ZFS_EVENT_CHANNEL "com.nexenta:zfs"
 420 #define ZFS_EC_STATUS   "status.zfs"
 421 #define ZPOOL_EC_STATUS "status.zpool"
 422 
 423 void zfs_event_post(const char *subclass,
 424     const char *operation, nvlist_t *ev_data);
 425 int zfs_ioc_set_prop_impl(char *name, nvlist_t *props,
 426     boolean_t received, nvlist_t **out_errors);
 427 
 428 
 429 #ifdef _KERNEL
 430 struct objset;
 431 struct zfsvfs;
 432 
 433 typedef struct zfs_creat {
 434         nvlist_t        *zct_zplprops;
 435         nvlist_t        *zct_props;
 436 } zfs_creat_t;
 437 
 438 extern dev_info_t *zfs_dip;
 439 
 440 extern int zfs_secpolicy_snapshot_perms(const char *, cred_t *);
 441 extern int zfs_secpolicy_rename_perms(const char *, const char *, cred_t *);
 442 extern int zfs_secpolicy_destroy_perms(const char *, cred_t *);
 443 extern int zfs_busy(void);
 444 extern void zfs_unmount_snap(const char *);
 445 extern void zfs_destroy_unmount_origin(const char *);
 446 extern int getzfsvfs_impl(struct objset *, struct zfsvfs **);
 447 extern int getzfsvfs(const char *, struct zfsvfs **);
 448 
 449 /*
 450  * ZFS minor numbers can refer to either a control device instance or
 451  * a zvol. Depending on the value of zss_type, zss_data points to either
 452  * a zvol_state_t or a zfs_onexit_t.
 453  */
 454 enum zfs_soft_state_type {
 455         ZSST_ZVOL,
 456         ZSST_CTLDEV
 457 };
 458 
 459 typedef struct zfs_soft_state {
 460         enum zfs_soft_state_type zss_type;
 461         void *zss_data;
 462 } zfs_soft_state_t;
 463 
 464 extern void *zfsdev_get_soft_state(minor_t minor,
 465     enum zfs_soft_state_type which);
 466 extern minor_t zfsdev_minor_alloc(void);
 467 
 468 extern void *zfsdev_state;
 469 extern kmutex_t zfsdev_state_lock;
 470 
 471 #endif  /* _KERNEL */
 472 
 473 #ifdef  __cplusplus
 474 }
 475 #endif
 476 
 477 #endif  /* _SYS_ZFS_IOCTL_H */