]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Add --offroot argument to mdmon
[thirdparty/mdadm.git] / mdadm.h
CommitLineData
64c4757e 1/*
9a9dab36 2 * mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e 3 *
e736b623 4 * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
64c4757e
NB
5 *
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 * Author: Neil Brown
e736b623 22 * Email: <neilb@suse.de>
64c4757e
NB
23 */
24
4ccdb956 25#define _GNU_SOURCE
c2c9bb6f 26#define _FILE_OFFSET_BITS 64
64c4757e 27#include <unistd.h>
e39b673e 28#if !defined(__dietlibc__) && !defined(__KLIBC__)
64c4757e 29extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
f783ca4f 30#else
fffdbe5e 31# if defined(__NO_STAT64) || __WORDSIZE != 32
f783ca4f
NB
32# define lseek64 lseek
33# endif
98c6faba 34#endif
64c4757e
NB
35
36#include <sys/types.h>
37#include <sys/stat.h>
38#include <stdlib.h>
39#include <time.h>
11018a45 40#include <sys/time.h>
64c4757e
NB
41#include <getopt.h>
42#include <fcntl.h>
43#include <stdio.h>
44#include <errno.h>
45#include <string.h>
773135f5 46#include <syslog.h>
b56c3630
NB
47#ifdef __dietlibc__
48#include <strings.h>
280a927d
NB
49/* dietlibc has deprecated random and srandom!! */
50#define random rand
51#define srandom srand
98c6faba
NB
52#endif
53
64c4757e
NB
54
55#include <linux/kdev_t.h>
e0d19036
NB
56/*#include <linux/fs.h> */
57#include <sys/mount.h>
58#include <asm/types.h>
64c4757e
NB
59#include <sys/ioctl.h>
60#define MD_MAJOR 9
dd0781e5 61#define MdpMinorShift 6
64c4757e 62
e0d19036 63#ifndef BLKGETSIZE64
98c6faba 64#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 65#endif
56eb10c0 66
c21e737b 67#define DEFAULT_CHUNK 512
c82f047c
NB
68#define DEFAULT_BITMAP_CHUNK 4096
69#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 70#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e 71
753cf905 72/* MAP_DIR should be somewhere that persists across the pivotroot
5d4d1b26 73 * from early boot to late boot.
7f5de63d 74 * Currently /dev seems to be the only option on most distros.
5d4d1b26 75 */
753cf905 76#ifndef MAP_DIR
7f5de63d 77#define MAP_DIR "/dev/.mdadm"
753cf905
DL
78#endif /* MAP_DIR */
79/* MAP_FILE is what we name the map file we put in MAP_DIR, in case you
80 * want something other than the default of "map"
81 */
82#ifndef MAP_FILE
83#define MAP_FILE "map"
84#endif /* MAP_FILE */
85/* MDMON_DIR is where pid and socket files used for communicating
86 * with mdmon normally live. It *should* be /var/run, but when
87 * mdmon is needed at early boot then it needs to write there prior
88 * to /var/run being mounted read/write, and it also then needs to
89 * persist beyond when /var/run is mounter read-only. So, to be
90 * safe, the default is somewhere that is read/write early in the
91 * boot process and stays up as long as possible during shutdown.
92 */
93#ifndef MDMON_DIR
7f5de63d 94#define MDMON_DIR "/dev/.mdadm/"
753cf905 95#endif /* MDMON_DIR */
5d4d1b26 96
403410eb
PC
97/* FAILED_SLOTS is where to save files storing recent removal of array
98 * member in order to allow future reuse of disk inserted in the same
99 * slot for array recovery
100 */
101#ifndef FAILED_SLOTS_DIR
102#define FAILED_SLOTS_DIR "/dev/.mdadm/failed-slots"
103#endif /* FAILED_SLOTS */
104
64c4757e 105#include "md_u.h"
e0d19036 106#include "md_p.h"
c82f047c 107#include "bitmap.h"
f7dd881f 108#include "msg.h"
64c4757e 109
1e0d770c 110#include <endian.h>
efd441d1
NB
111/* Redhat don't like to #include <asm/byteorder.h>, and
112 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
113 * and there is no standard conversion function so... */
0ae03b8a
NB
114/* And dietlibc doesn't think byteswap is ok, so.. */
115/* #include <byteswap.h> */
116#define bswap_16(x) (((x) & 0x00ffU) << 8 | \
117 ((x) & 0xff00U) >> 8)
118#define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
119 ((x) & 0xff000000U) >> 24 | \
120 ((x) & 0x0000ff00U) << 8 | \
121 ((x) & 0x00ff0000U) >> 8)
122#define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
123 ((x) & 0xff00000000000000ULL) >> 56 | \
124 ((x) & 0x000000000000ff00ULL) << 40 | \
125 ((x) & 0x00ff000000000000ULL) >> 40 | \
126 ((x) & 0x0000000000ff0000ULL) << 24 | \
127 ((x) & 0x0000ff0000000000ULL) >> 24 | \
128 ((x) & 0x00000000ff000000ULL) << 8 | \
129 ((x) & 0x000000ff00000000ULL) >> 8)
130
e39b673e 131#if !defined(__KLIBC__)
efd441d1
NB
132#if BYTE_ORDER == LITTLE_ENDIAN
133#define __cpu_to_le16(_x) (_x)
134#define __cpu_to_le32(_x) (_x)
135#define __cpu_to_le64(_x) (_x)
136#define __le16_to_cpu(_x) (_x)
137#define __le32_to_cpu(_x) (_x)
138#define __le64_to_cpu(_x) (_x)
974e620d
NB
139
140#define __cpu_to_be16(_x) bswap_16(_x)
141#define __cpu_to_be32(_x) bswap_32(_x)
142#define __cpu_to_be64(_x) bswap_64(_x)
143#define __be16_to_cpu(_x) bswap_16(_x)
144#define __be32_to_cpu(_x) bswap_32(_x)
145#define __be64_to_cpu(_x) bswap_64(_x)
efd441d1
NB
146#elif BYTE_ORDER == BIG_ENDIAN
147#define __cpu_to_le16(_x) bswap_16(_x)
148#define __cpu_to_le32(_x) bswap_32(_x)
149#define __cpu_to_le64(_x) bswap_64(_x)
150#define __le16_to_cpu(_x) bswap_16(_x)
151#define __le32_to_cpu(_x) bswap_32(_x)
152#define __le64_to_cpu(_x) bswap_64(_x)
974e620d
NB
153
154#define __cpu_to_be16(_x) (_x)
155#define __cpu_to_be32(_x) (_x)
156#define __cpu_to_be64(_x) (_x)
157#define __be16_to_cpu(_x) (_x)
158#define __be32_to_cpu(_x) (_x)
159#define __be64_to_cpu(_x) (_x)
1e0d770c
NB
160#else
161# error "unknown endianness."
162#endif
e39b673e 163#endif /* __KLIBC__ */
1e0d770c
NB
164
165
1e5c6983
DW
166/*
167 * min()/max()/clamp() macros that also do
168 * strict type-checking.. See the
169 * "unnecessary" pointer comparison.
170 */
171#define min(x, y) ({ \
172 typeof(x) _min1 = (x); \
173 typeof(y) _min2 = (y); \
174 (void) (&_min1 == &_min2); \
175 _min1 < _min2 ? _min1 : _min2; })
176
177#define max(x, y) ({ \
178 typeof(x) _max1 = (x); \
179 typeof(y) _max2 = (y); \
180 (void) (&_max1 == &_max2); \
181 _max1 > _max2 ? _max1 : _max2; })
1e0d770c 182
4b1ac34b
NB
183/* general information that might be extracted from a superblock */
184struct mdinfo {
185 mdu_array_info_t array;
186 mdu_disk_info_t disk;
187 __u64 events;
fbf8a0b7 188 int uuid[4];
31317663 189 char name[33];
353632d9 190 unsigned long long data_offset;
598f0d58
NB
191 unsigned long long component_size; /* same as array.size, except in
192 * sectors and up to 64bits.
193 */
da9b4a62
DW
194 unsigned long long custom_array_size; /* size for non-default sized
195 * arrays (in sectors)
196 */
81219e70
LM
197#define NO_RESHAPE 0
198#define VOLUME_RESHAPE 1
199#define CONTAINER_RESHAPE 2
353632d9
NB
200 int reshape_active;
201 unsigned long long reshape_progress;
6e75048b
AK
202 int recovery_blocked; /* for external metadata it
203 * indicates that there is
204 * reshape in progress in
205 * container,
206 * for native metadata it is
207 * reshape_active field mirror
208 */
e1516be1
DW
209 union {
210 unsigned long long resync_start; /* per-array resync position */
211 unsigned long long recovery_start; /* per-device rebuild position */
b7528a20 212 #define MaxSector (~0ULL) /* resync/recovery complete position */
e1516be1 213 };
c0c1acd6 214 unsigned long bitmap_offset; /* 0 == none, 1 == a file */
a67dd8cc 215 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 216 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 217 int errors;
f21e18ca 218 unsigned long cache_size; /* size of raid456 stripe cache*/
7e0f6979
NB
219 int mismatch_cnt;
220 char text_version[50];
dd15dc4a
NB
221
222 int container_member; /* for assembling external-metatdata arrays
223 * This is to be used internally by metadata
224 * handler only */
97b4d0e9
DW
225 int container_enough; /* flag external handlers can set to
226 * indicate that subarrays have not enough (-1),
227 * enough to start (0), or all expected disks (1) */
06c7f68e 228 char sys_name[20];
7e0f6979 229 struct mdinfo *devs;
06c7f68e 230 struct mdinfo *next;
549e9569
NB
231
232 /* Device info for mdmon: */
e1516be1 233 int recovery_fd;
549e9569 234 int state_fd;
8d45d196
DW
235 #define DS_FAULTY 1
236 #define DS_INSYNC 2
237 #define DS_WRITE_MOSTLY 4
238 #define DS_SPARE 8
239 #define DS_BLOCKED 16
240 #define DS_REMOVE 1024
92967543 241 #define DS_UNBLOCK 2048
549e9569
NB
242 int prev_state, curr_state, next_state;
243
4b1ac34b
NB
244};
245
5bbb4842
NB
246struct createinfo {
247 int uid;
248 int gid;
249 int autof;
250 int mode;
38098016 251 int symlinks;
058574b1 252 struct supertype *supertype;
5bbb4842
NB
253};
254
9a9dab36 255#define Name "mdadm"
682c7051 256
e0d19036
NB
257enum mode {
258 ASSEMBLE=1,
259 BUILD,
260 CREATE,
261 MANAGE,
262 MISC,
263 MONITOR,
dd0781e5 264 GROW,
8382f19b 265 INCREMENTAL,
1f48664b 266 AUTODETECT,
e0d19036
NB
267};
268
64c4757e 269extern char short_options[];
024768c4 270extern char short_bitmap_options[];
c06487ce 271extern char short_bitmap_auto_options[];
64c4757e 272extern struct option long_options[];
56eedc1a 273extern char Version[], Usage[], Help[], OptionHelp[],
dd0781e5 274 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 275 Help_incr[],
e0d19036 276 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 277
997aed5d 278/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 279 * numbers later than any 'short' character option.
997aed5d
NB
280 */
281enum special_options {
1c7a808c 282 AssumeClean = 300,
997aed5d
NB
283 BitmapChunk,
284 WriteBehind,
285 ReAdd,
286 NoDegraded,
287 Sparc22,
1c7a808c 288 BackupFile,
997aed5d 289 HomeHost,
589395d6 290 AutoHomeHost,
38098016 291 Symlinks,
1f48664b 292 AutoDetect,
1770662b 293 Waitclean,
4cce4069 294 DetailPlatform,
33414a01 295 KillSubarray,
1c7a808c 296 UpdateSubarray,
edde9560 297 IncrementalPath,
1c7a808c
N
298 NoSharing,
299 HelpOptions,
300 Brief,
301 ManageOpt,
302 Add,
303 Remove,
304 Fail,
305 MiscOpt,
306 WaitOpt,
307 ConfigFile,
308 ChunkSize,
309 WriteMostly,
310 Layout,
311 Auto,
312 Force,
313 SuperMinor,
314 EMail,
315 ProgramOpt,
316 Increment,
317 Fork,
318 Bitmap,
319 RebuildMapOpt,
87f26d14 320 InvalidBackup,
20b60dcd 321 UdevRules,
b76b30e0 322 FreezeReshape,
2dddadb0 323 Continue,
08ca2adf 324 OffRootOpt,
997aed5d
NB
325};
326
64c4757e 327/* structures read from config file */
52826846
NB
328/* List of mddevice names and identifiers
329 * Identifiers can be:
330 * uuid=128-hex-uuid
331 * super-minor=decimal-minor-number-from-superblock
332 * devices=comma,separated,list,of,device,names,with,wildcards
333 *
334 * If multiple fields are present, the intersection of all matching
335 * devices is considered
336 */
98c6faba 337#define UnSet (0xfffe)
fa56eddb 338struct mddev_ident {
dd0781e5 339 char *devname;
aba69144 340
dd0781e5 341 int uuid_set;
3fa06e9d 342 int uuid[4];
947fd4dd 343 char name[33];
52826846 344
f21e18ca 345 int super_minor;
52826846 346
dd0781e5 347 char *devices; /* comma separated list of device
52826846
NB
348 * names with wild cards
349 */
dd0781e5 350 int level;
f21e18ca
N
351 int raid_disks;
352 int spare_disks;
82d9eba6 353 struct supertype *st;
dd0781e5
NB
354 int autof; /* 1 for normal, 2 for partitioned */
355 char *spare_group;
7ef02d01 356 char *bitmap_file;
c82f047c 357 int bitmap_fd;
dd0781e5 358
1771a6e2
N
359 char *container; /* /dev/whatever name of container, or
360 * uuid of container. You would expect
361 * this to be the 'devname' or UUID
dbb44303
N
362 * of some other entry.
363 */
364 char *member; /* subarray within a container */
365
fa56eddb 366 struct mddev_ident *next;
b179246f
N
367 union {
368 /* fields needed by different users of this structure */
369 int assembled; /* set when assembly succeeds */
370 };
fa56eddb 371};
64c4757e
NB
372
373/* List of device names - wildcards expanded */
a655e550 374struct mddev_dev {
64c4757e 375 char *devname;
1c7a808c 376 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
cd29a5c8
NB
377 * Not set for names read from .config
378 */
b3d31955 379 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
fe80f49b 380 char re_add;
811e6cbe 381 char used; /* set when used */
a655e550
N
382 struct mddev_dev *next;
383};
64c4757e 384
682c7051
NB
385typedef struct mapping {
386 char *name;
387 int num;
388} mapping_t;
389
e0d19036
NB
390
391struct mdstat_ent {
392 char *dev;
393 int devnum;
394 int active;
395 char *level;
396 char *pattern; /* U or up, _ for down */
397 int percent; /* -1 if no resync */
f94c116f 398 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
399 int devcnt;
400 int raid_disks;
549e9569 401 char * metadata_version;
3b57c466
N
402 struct dev_member {
403 char *name;
404 struct dev_member *next;
405 } *members;
e0d19036
NB
406 struct mdstat_ent *next;
407};
408
22a88995 409extern struct mdstat_ent *mdstat_read(int hold, int start);
e0d19036 410extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 411extern void mdstat_wait(int seconds);
58a4ba2a 412extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
8382f19b 413extern int mddev_busy(int devnum);
3b57c466 414extern struct mdstat_ent *mdstat_by_component(char *name);
78b10e66 415extern struct mdstat_ent *mdstat_by_subdev(char *subdev, int container);
8382f19b
NB
416
417struct map_ent {
418 struct map_ent *next;
419 int devnum;
1522c538 420 char metadata[20];
8382f19b 421 int uuid[4];
195254b8 422 int bad;
8382f19b
NB
423 char *path;
424};
1522c538 425extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
8382f19b 426 int uuid[4], char *path);
4eb26970 427extern void map_remove(struct map_ent **map, int devnum);
8382f19b 428extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
f2e55ecc
N
429extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
430extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
431extern void map_read(struct map_ent **melp);
432extern int map_write(struct map_ent *mel);
433extern void map_delete(struct map_ent **mapp, int devnum);
434extern void map_free(struct map_ent *map);
435extern void map_add(struct map_ent **melp,
1522c538 436 int devnum, char *metadata, int uuid[4], char *path);
ad5bc697
N
437extern int map_lock(struct map_ent **melp);
438extern void map_unlock(struct map_ent **melp);
cc700db3 439extern void map_fork(void);
e0d19036 440
e86c9dd6 441/* various details can be requested */
dab4a513
DW
442enum sysfs_read_flags {
443 GET_LEVEL = (1 << 0),
444 GET_LAYOUT = (1 << 1),
445 GET_COMPONENT = (1 << 2),
446 GET_CHUNK = (1 << 3),
447 GET_CACHE = (1 << 4),
448 GET_MISMATCH = (1 << 5),
449 GET_VERSION = (1 << 6),
450 GET_DISKS = (1 << 7),
451 GET_DEGRADED = (1 << 8),
452 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
453 GET_BITMAP_LOCATION = (1 << 10),
454
455 GET_DEVS = (1 << 20), /* gets role, major, minor */
456 GET_OFFSET = (1 << 21),
457 GET_SIZE = (1 << 22),
458 GET_STATE = (1 << 23),
459 GET_ERROR = (1 << 24),
dab4a513 460};
e86c9dd6
NB
461
462/* If fd >= 0, get the array it is open on,
463 * else use devnum. >=0 -> major9. <0.....
464 */
549e9569 465extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 466extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
467extern void sysfs_free(struct mdinfo *sra);
468extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
469extern int sysfs_attr_match(const char *attr, const char *str);
470extern int sysfs_match_word(const char *word, char **list);
7e0f6979 471extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 472 char *name, char *val);
7e0f6979 473extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 474 char *name, unsigned long long val);
97590376 475extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
476extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
477 char *name);
478extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 479extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 480 char *name, unsigned long long *val);
7236ee7a 481extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
482extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
483 char *name);
93ecfa01
N
484extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
485 char *name, char *val, int size);
8ed3e5e1 486extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 487extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 488extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 489extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 490extern int sysfs_unique_holder(int devnum, long rdev);
bc77ed53 491extern int sysfs_freeze_array(struct mdinfo *sra);
c69b251b 492extern int load_sys(char *path, char *buf);
130994cb
AK
493extern int reshape_prepare_fdlist(char *devname,
494 struct mdinfo *sra,
495 int raid_disks,
496 int nrdisks,
497 unsigned long blocks,
498 char *backup_file,
499 int *fdlist,
500 unsigned long long *offsets);
501extern void reshape_free_fdlist(int *fdlist,
502 unsigned long long *offsets,
503 int size);
e6e9d47b
AK
504extern int reshape_open_backup_file(char *backup,
505 int fd,
506 char *devname,
507 long blocks,
508 int *fdlist,
a93f87ee
N
509 unsigned long long *offsets,
510 int restart);
1c009fc2
AK
511extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
512 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
513
514extern int save_stripes(int *source, unsigned long long *offsets,
515 int raid_disks, int chunk_size, int level, int layout,
516 int nwrites, int *dest,
a6288483
N
517 unsigned long long start, unsigned long long length,
518 char *buf);
353632d9
NB
519extern int restore_stripes(int *dest, unsigned long long *offsets,
520 int raid_disks, int chunk_size, int level, int layout,
521 int source, unsigned long long read_offset,
2fcb75ae
AK
522 unsigned long long start, unsigned long long length,
523 char *src_buf);
e86c9dd6 524
52826846
NB
525#ifndef Sendmail
526#define Sendmail "/usr/lib/sendmail -t"
527#endif
528
773135f5
NB
529#define SYSLOG_FACILITY LOG_DAEMON
530
682c7051
NB
531extern char *map_num(mapping_t *map, int num);
532extern int map_name(mapping_t *map, char *name);
b640a252 533extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 534
16c6fa80 535extern char *map_dev(int major, int minor, int create);
64c4757e 536
549e9569 537struct active_array;
2e735d19 538struct metadata_update;
64c4757e 539
999b4972
N
540
541/* 'struct reshape' records the intermediate states
542 * a general reshape.
543 * The starting geometry is converted to the 'before' geometry
544 * by at most an atomic level change. They could be the same.
545 * Similarly the 'after' geometry is converted to the final
546 * geometry by at most a level change.
547 * Note that 'before' and 'after' must have the same level.
548 * 'blocks' is the minimum number of sectors for a reshape unit.
549 * This will be a multiple of the stripe size in each of the
550 * 'before' and 'after' geometries.
551 * If 'blocks' is 0, no restriping is necessary.
552 */
553struct reshape {
554 int level;
555 int parity; /* number of parity blocks/devices */
556 struct {
557 int layout;
558 int data_disks;
559 } before, after;
560 unsigned long long backup_blocks;
561 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
562 unsigned long long new_size; /* New size of array in sectors */
563};
564
6adfd3af
NB
565/* A superswitch provides entry point the a metadata handler.
566 *
567 * The super_switch primarily operates on some "metadata" that
568 * is accessed via the 'supertype'.
569 * This metadata has one of three possible sources.
570 * 1/ It is read from a single device. In this case it may not completely
571 * describe the array or arrays as some information might be on other
572 * devices.
573 * 2/ It is read from all devices in a container. In this case all
574 * information is present.
575 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
576 * be complete once enough ->add_to_super calls have completed.
577 *
578 * When creating an array inside a container, the metadata will be
579 * formed by a combination of 2 and 3. The metadata or the array is read,
580 * then new information is added.
581 *
582 * The metadata must sometimes have a concept of a 'current' array
583 * and a 'current' device.
584 * The 'current' array is set by init_super to be the newly created array,
585 * or is set by super_by_fd when it finds it is looking at an array inside
586 * a container.
587 *
588 * The 'current' device is either the device that the metadata was read from
589 * in case 1, or the last device added by add_to_super in case 3.
590 * Case 2 does not identify a 'current' device.
591 */
f9ce90ba 592extern struct superswitch {
6adfd3af
NB
593
594 /* Used to report details of metadata read from a component
595 * device. ->load_super has been called.
596 */
3da92f27 597 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 598 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 599 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 600 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
601
602 /* Used to report details of an active array.
603 * ->load_super was possibly given a 'component' string.
604 */
3da92f27 605 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 606 void (*brief_detail_super)(struct supertype *st);
0d726f17 607 void (*export_detail_super)(struct supertype *st);
6adfd3af 608
4cce4069 609 /* Optional: platform hardware / firmware details */
5615172f 610 int (*detail_platform)(int verbose, int enumerate_only);
4cce4069 611
6adfd3af
NB
612 /* Used:
613 * to get uuid to storing in bitmap metadata
614 * and 'reshape' backup-data metadata
615 * To see if a device is being re-added to an array it was part of.
616 */
3da92f27 617 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 618
8592f29d 619 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
620 * the container, or about an individual array within the container.
621 * The determination is made either by:
622 * load_super being given a 'component' string.
623 * validate_geometry determining what to create.
d2ca6449
NB
624 * The info includes both array information and device information.
625 * The particular device should be:
626 * The last device added by add_to_super
627 * The device the metadata was loaded from by load_super
a5d85af7
N
628 * If 'map' is present, then it is an array raid_disks long
629 * (raid_disk must already be set and correct) and it is filled
630 * with 1 for slots that are thought to be active and 0 for slots which
631 * appear to be failed/missing.
95eeceeb 632 * *info is zeroed out before data is added.
6adfd3af 633 */
a5d85af7 634 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 635 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 636 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 637 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 638 */
3da92f27 639 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
640
641 /* Make one of several generic modifications to metadata
642 * prior to assembly (or other times).
643 * sparc2.2 - first bug in early 0.90 metadata
644 * super-minor - change name of 0.90 metadata
645 * summaries - 'correct' any redundant data
646 * resync - mark array as dirty to trigger a resync.
647 * uuid - set new uuid - only 0.90 or 1.x
648 * name - change the name of the array (where supported)
649 * homehost - change which host this array is tied to.
650 * devicesize - If metadata is at start of device, change recorded
651 * device size to match actual device size
652 * byteorder - swap bytes for 0.90 metadata
653 *
654 * force-one - mark that device as uptodate, not old or failed.
655 * force-array - mark array as clean if it would not otherwise
656 * assemble
657 * assemble - not sure how this is different from force-one...
658 * linear-grow-new - add a new device to a linear array, but don't
659 * change the size: so superblock still matches
660 * linear-grow-update - now change the size of the array.
16715c01
DL
661 * writemostly - set the WriteMostly1 bit in the superblock devflags
662 * readwrite - clear the WriteMostly1 bit in the superblock devflags
6adfd3af 663 */
68c7d6d7 664 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 665 char *update,
e5eac01f
NB
666 char *devname, int verbose,
667 int uuid_set, char *homehost);
6adfd3af
NB
668
669 /* Create new metadata for new array as described. This could
670 * be a new container, or an array in a pre-existing container.
671 * Also used to zero metadata prior to writing it to invalidate old
672 * metadata.
673 */
3da92f27
NB
674 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
675 unsigned long long size, char *name,
676 char *homehost, int *uuid);
6adfd3af
NB
677
678 /* update the metadata to include new device, either at create or
679 * when hot-adding a spare.
680 */
f20c3968 681 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
1a64be56
LM
682 int fd, char *devname);
683 /* update the metadata to delete a device,
684 * when hot-removing.
685 */
686 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
687
688 /* Write metadata to one device when fixing problems or adding
689 * a new device.
690 */
3da92f27 691 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
692
693 /* Write all metadata for this array.
694 */
111d01fc 695 int (*write_init_super)(struct supertype *st);
64557c33 696 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 697 int (*load_super)(struct supertype *st, int fd, char *devname);
2b959fbf 698 int (*load_container)(struct supertype *st, int fd, char *devname);
82d9eba6 699 struct supertype * (*match_metadata_desc)(char *arg);
1bf4e2d9 700 __u64 (*avail_size)(struct supertype *st, __u64 size);
80e7f8c3 701 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
3da92f27 702 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 703 int delay, int write_behind,
f9c25f1d 704 unsigned long long size, int may_change, int major);
3da92f27
NB
705 void (*locate_bitmap)(struct supertype *st, int fd);
706 int (*write_bitmap)(struct supertype *st, int fd);
707 void (*free_super)(struct supertype *st);
78e44928
NB
708
709 /* validate_geometry is called with an st returned by
710 * match_metadata_desc.
711 * It should check that the geometry described in compatible with
712 * the metadata type. It will be called repeatedly as devices
713 * added to validate changing size and new devices. If there are
714 * inter-device dependencies, it should record sufficient details
715 * so these can be validated.
30f58b22 716 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
717 * Return value is:
718 * 1: everything is OK
719 * 0: not OK for some reason - if 'verbose', then error was reported.
720 * -1: st->sb was NULL, 'subdev' is a member of a container of this
721 * types, but array is not acceptable for some reason
722 * message was reported even if verbose is 0.
78e44928 723 */
17f25ca6
NB
724 int (*validate_geometry)(struct supertype *st, int level, int layout,
725 int raiddisks,
c21e737b 726 int *chunk, unsigned long long size,
2c514b71
NB
727 char *subdev, unsigned long long *freesize,
728 int verbose);
598f0d58 729
00bbdbda 730 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
731 /* query the supertype for default geometry */
732 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
733 /* Permit subarray's to be deleted from inactive containers */
734 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 735 /* Permit subarray's to be modified */
a951a4f7 736 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 737 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
738 /* Check if reshape is supported for this external format.
739 * st is obtained from super_by_fd() where st->subarray[0] is
740 * initialized to indicate if reshape is being performed at the
741 * container or subarray level
742 */
743 int (*reshape_super)(struct supertype *st, long long size, int level,
744 int layout, int chunksize, int raid_disks,
41784c88
AK
745 int delta_disks, char *backup, char *dev,
746 int verbose); /* optional */
999b4972
N
747 int (*manage_reshape)( /* optional */
748 int afd, struct mdinfo *sra, struct reshape *reshape,
749 struct supertype *st, unsigned long blocks,
750 int *fds, unsigned long long *offsets,
751 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 752
549e9569 753/* for mdmon */
cba0191b
NB
754 int (*open_new)(struct supertype *c, struct active_array *a,
755 char *inst);
ed9d66aa
NB
756
757 /* Tell the metadata handler the current state of the array.
758 * This covers whether it is known to be consistent (no pending writes)
01f157d7 759 * and how far along a resync is known to have progressed
ed9d66aa
NB
760 * (in a->resync_start).
761 * resync status is really irrelevant if the array is not consistent,
762 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
763 * If 'consistent' is '2', then the array can mark it dirty if a
764 * resync/recovery/whatever is required, or leave it clean if not.
765 * Return value is 0 dirty (not consistent) and 1 if clean.
766 * it is only really important if consistent is passed in as '2'.
4e5528c6 767 */
01f157d7 768 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
769
770 /* When the state of a device might have changed, we call set_disk to
771 * tell the metadata what the current state is.
772 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
773 * transitions.
774 * set_disk might be called when the state of the particular disk has
775 * not in fact changed.
776 */
8d45d196 777 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
778 void (*sync_metadata)(struct supertype *st);
779 void (*process_update)(struct supertype *st,
780 struct metadata_update *update);
edd8d13c
NB
781 void (*prepare_update)(struct supertype *st,
782 struct metadata_update *update);
549e9569 783
6c3fb95c
NB
784 /* activate_spare will check if the array is degraded and, if it
785 * is, try to find some spare space in the container.
786 * On success, it add appropriate updates (For process_update) to
787 * to the 'updates' list and returns a list of 'mdinfo' identifying
788 * the device, or devices as there might be multiple missing
789 * devices and multiple spares available.
790 */
791 struct mdinfo *(*activate_spare)(struct active_array *a,
792 struct metadata_update **updates);
2cda7640
ML
793 /*
794 * Return statically allocated string that represents metadata specific
795 * controller domain of the disk. The domain is used in disk domain
796 * matching functions. Disks belong to the same domain if the they have
797 * the same domain from mdadm.conf and belong the same metadata domain.
798 * Returning NULL or not providing this handler means that metadata
799 * does not distinguish the differences between disks that belong to
800 * different controllers. They are in the domain specified by
801 * configuration file (mdadm.conf).
802 * In case when the metadata has the notion of domains based on disk
803 * it shall return NULL for disks that do not belong to the controller
804 * the supported domains. Such disks will form another domain and won't
805 * be mixed with supported ones.
806 */
807 const char *(*get_disk_controller_domain)(const char *path);
549e9569 808
ba53ea59
AK
809 /* for external backup area */
810 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
811
f277ce36 812 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 813 int external;
4cce4069 814 const char *name; /* canonical metadata name */
0f22b998 815} *superlist[];
82d9eba6 816
0f22b998
N
817extern struct superswitch super0, super1;
818extern struct superswitch super_imsm, super_ddf;
0592faeb 819extern struct superswitch mbr, gpt;
cdddbdbc 820
edd8d13c
NB
821struct metadata_update {
822 int len;
823 char *buf;
824 void *space; /* allocated space that monitor will use */
cb23f1f4
N
825 void **space_list; /* list of allocated spaces that monitor can
826 * use or that it returned.
827 */
edd8d13c
NB
828 struct metadata_update *next;
829};
830
6adfd3af
NB
831/* A supertype holds a particular collection of metadata.
832 * It identifies the metadata type by the superswitch, and the particular
833 * sub-version of that metadata type.
834 * metadata read in or created is stored in 'sb' and 'info'.
835 * There are also fields used by mdmon to track containers.
836 *
352452c3
N
837 * A supertype may refer to:
838 * Just an array, possibly in a container
839 * A container, not identifying any particular array
840 * Info read from just one device, not yet fully describing the array/container.
841 *
842 *
6adfd3af
NB
843 * A supertype is created by:
844 * super_by_fd
845 * guess_super
846 * dup_super
847 */
82d9eba6
NB
848struct supertype {
849 struct superswitch *ss;
850 int minor_version;
ea329559 851 int max_devs;
2f6079dc 852 int container_dev; /* devnum of container */
64557c33 853 void *sb;
111d01fc 854 void *info;
df3346e6
LM
855 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
856 HW/firmware related incompatability to load metadata.
857 Used when examining metadata to display content of disk
858 when user has no hw/firmare compatible system.
859 */
edd8d13c
NB
860 struct metadata_update *updates;
861 struct metadata_update **update_tail;
862
549e9569
NB
863 /* extra stuff used by mdmon */
864 struct active_array *arrays;
549e9569 865 int sock; /* listen to external programs */
549e9569
NB
866 int devnum;
867 char *devname; /* e.g. md0. This appears in metadata_verison:
868 * external:/md0/12
869 */
870 int devcnt;
68226a80 871 int retry_soon;
549e9569
NB
872
873 struct mdinfo *devs;
874
82d9eba6 875};
f9ce90ba 876
4725bc31 877extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
878enum guess_types { guess_any, guess_array, guess_partitions };
879extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
880static inline struct supertype *guess_super(int fd) {
881 return guess_super_type(fd, guess_any);
882}
3da92f27 883extern struct supertype *dup_super(struct supertype *st);
beae1dfe 884extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 885extern int must_be_container(int fd);
0f0749ad 886extern int dev_size_from_id(dev_t id, unsigned long long *size);
8382f19b
NB
887extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
888 mdu_disk_info_t *disk);
a7c6e3fb 889void wait_for(char *dev, int fd);
f9ce90ba 890
5527fc74
N
891/*
892 * Data structures for policy management.
893 * Each device can have a policy structure that lists
894 * various name/value pairs each possibly with a metadata associated.
895 * The policy list is sorted by name/value/metadata
896 */
897struct dev_policy {
898 struct dev_policy *next;
899 char *name; /* None of these strings are allocated. They are
900 * all just references to strings which are known
901 * to exist elsewhere.
902 * name and metadata can be compared by address equality.
903 */
904 const char *metadata;
2cda7640 905 const char *value;
5527fc74
N
906};
907
4e8d9f0a 908extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
909
910/* iterate over the sublist starting at list, having the same
911 * 'name' as 'list', and matching the given metadata (Where
912 * NULL matches anything
913 */
914#define pol_for_each(item, list, _metadata) \
915 for (item = list; \
916 item && item->name == list->name; \
917 item = item->next) \
918 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
919 ; else
920
921/*
922 * policy records read from mdadm are largely just name-value pairs.
923 * The names are constants, not strdupped
924 */
925struct pol_rule {
926 struct pol_rule *next;
927 char *type; /* rule_policy or rule_part */
928 struct rule {
929 struct rule *next;
930 char *name;
931 char *value;
932 char *dups; /* duplicates of 'value' with a partNN appended */
933 } *rule;
934};
935
936extern char rule_policy[], rule_part[];
937extern char rule_path[], rule_type[];
938extern char type_part[], type_disk[];
939
940extern void policyline(char *line, char *type);
4e8d9f0a 941extern void policy_add(char *type, ...);
5527fc74
N
942extern void policy_free(void);
943
73c9c47c 944extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 945extern struct dev_policy *disk_policy(struct mdinfo *disk);
4e8d9f0a 946extern struct dev_policy *devnum_policy(int dev);
5527fc74
N
947extern void dev_policy_free(struct dev_policy *p);
948
e78dda3b
N
949//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
950extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
951extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
952
e3bb5f14
N
953enum policy_action {
954 act_default,
955 act_include,
956 act_re_add,
d2db3045
N
957 act_spare, /* This only applies to bare devices */
958 act_spare_same_slot, /* this allows non-bare devices,
959 * but only if recent removal */
960 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
961 act_err
962};
963
964extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
965 enum policy_action want);
966extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
967 enum policy_action want);
f5f12c84
N
968
969struct domainlist {
970 struct domainlist *next;
2cda7640 971 const char *dom;
f5f12c84
N
972};
973
974extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
975 const char *metadata);
976extern struct domainlist *domain_from_array(struct mdinfo *mdi,
977 const char *metadata);
e78dda3b
N
978extern void domainlist_add_dev(struct domainlist **dom, int devnum,
979 const char *metadata);
f5f12c84
N
980extern void domain_free(struct domainlist *dl);
981extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
982 const char *metadata);
e78dda3b 983void domain_add(struct domainlist **domp, char *domain);
f5f12c84 984
403410eb
PC
985extern void policy_save_path(char *id_path, struct map_ent *array);
986extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
987
838acbc2
NB
988#if __GNUC__ < 3
989struct stat64;
990#endif
991
45e878bb
NB
992#define HAVE_NFTW we assume
993#define HAVE_FTW
994
2df1f269 995#ifdef __UCLIBC__
45e878bb 996# include <features.h>
2df1f269
BRF
997# ifndef __UCLIBC_HAS_LFS__
998# define lseek64 lseek
999# endif
45e878bb
NB
1000# ifndef __UCLIBC_HAS_FTW__
1001# undef HAVE_FTW
1002# undef HAVE_NFTW
1003# endif
1004#endif
1005
1006#ifdef __dietlibc__
1007# undef HAVE_NFTW
1008#endif
1009
e39b673e 1010#if defined(__KLIBC__)
1011# undef HAVE_NFTW
1012# undef HAVE_FTW
1013#endif
1014
45e878bb 1015#ifndef HAVE_NFTW
173fc515 1016# define FTW_PHYS 1
45e878bb
NB
1017# ifndef HAVE_FTW
1018 struct FTW {};
173fc515
NB
1019# endif
1020#endif
1021
45e878bb
NB
1022#ifdef HAVE_FTW
1023# include <ftw.h>
1024#endif
1025
173fc515
NB
1026extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1027
f9ce90ba 1028
64c4757e 1029extern int Manage_ro(char *devname, int fd, int readonly);
91f068bf 1030extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
dd0781e5 1031extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
64c4757e 1032extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1033 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1034 char *update, int force);
1f48664b 1035extern int autodetect(void);
e5329c37 1036extern int Grow_Add_device(char *devname, int fd, char *newdev);
8fac0577 1037extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
06b0d786 1038extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
e86c9dd6 1039 long long size,
691a36b7 1040 int level, char *layout_str, int chunksize, int raid_disks,
e2e53a2d 1041 struct mddev_dev *devlist,
ce52f92f 1042 int assume_clean, int force);
06b0d786 1043extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1044 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1045extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1046 struct mdinfo *info, char *backup_file,
1047 int freeze_reshape);
1048
3f54bd62
AK
1049extern int restore_backup(struct supertype *st,
1050 struct mdinfo *content,
1051 int working_disks,
1052 int spares,
1053 char *backup_file,
1054 int verbose);
2dddadb0
AK
1055extern int Grow_continue_command(char *devname, int fd,
1056 char *backup_file, int verbose);
64c4757e 1057
7f91af49 1058extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1059 struct mddev_ident *ident,
87f26d14
N
1060 struct mddev_dev *devlist,
1061 char *backup_file, int invalid_backup,
64c4757e 1062 int readonly, int runstop,
0ac91628 1063 char *update, char *homehost, int require_homehost,
b76b30e0 1064 int verbose, int force, int freeze_reshape);
64c4757e 1065
7f91af49 1066extern int Build(char *mddev, int chunk, int level, int layout,
a655e550 1067 int raiddisks, struct mddev_dev *devlist, int assume_clean,
7f91af49 1068 char *bitmap_file, int bitmap_chunk, int write_behind,
83208785 1069 int delay, int verbose, int autof, unsigned long long size);
64c4757e
NB
1070
1071
7f91af49 1072extern int Create(struct supertype *st, char *mddev,
5dd497ee 1073 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
3d3dd91e 1074 char *name, char *homehost, int *uuid,
a655e550 1075 int subdevs, struct mddev_dev *devlist,
47d79ef8 1076 int runstop, int verbose, int force, int assume_clean,
7f91af49 1077 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
64c4757e 1078
54bad364 1079extern int Detail(char *dev, int brief, int export, int test, char *homehost);
4cce4069 1080extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
e0d19036 1081extern int Query(char *dev);
a655e550 1082extern int Examine(struct mddev_dev *devlist, int brief, int export, int scan,
0d726f17 1083 int SparcAdjust, struct supertype *forcest, char *homehost);
a655e550 1084extern int Monitor(struct mddev_dev *devlist,
e0d19036 1085 char *mailaddr, char *alert_cmd,
aa88f531 1086 int period, int daemonise, int scan, int oneshot,
edde9560
AC
1087 int dosyslog, int test, char *pidfile, int increments,
1088 int share);
64c4757e 1089
9277cc77 1090extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
33414a01 1091extern int Kill_subarray(char *dev, char *subarray, int quiet);
fa56eddb 1092extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1093extern int Wait(char *dev);
9f1da824 1094extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1095
8382f19b 1096extern int Incremental(char *devname, int verbose, int runstop,
0ac91628 1097 struct supertype *st, char *homehost, int require_homehost,
b76b30e0 1098 int autof, int freeze_reshape);
8382f19b
NB
1099extern void RebuildMap(void);
1100extern int IncrementalScan(int verbose);
950bc344 1101extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1102extern int CreateBitmap(char *filename, int force, char uuid[16],
1103 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1104 unsigned long write_behind,
dcec9ee5
NB
1105 unsigned long long array_size,
1106 int major);
55935d51 1107extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1108extern int Write_rules(char *rule_name);
bf4fb153 1109extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1110extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 1111
64c4757e 1112extern int md_get_version(int fd);
e0d19036 1113extern int get_linux_version(void);
bc77ed53 1114extern int mdadm_version(char *version);
84e11361 1115extern long long parse_size(char *size);
64c4757e 1116extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1117extern int parse_layout_10(char *layout);
1118extern int parse_layout_faulty(char *layout);
682c7051
NB
1119extern int check_ext2(int fd, char *name);
1120extern int check_reiser(int fd, char *name);
1121extern int check_raid(int fd, char *name);
53ed6ac3
KW
1122extern int check_partitions(int fd, char *dname,
1123 unsigned long long freesize,
1124 unsigned long long size);
64c4757e 1125
dd0781e5 1126extern int get_mdp_major(void);
8b0dabea 1127extern int dev_open(char *dev, int flags);
e8a70c89 1128extern int open_dev(int devnum);
d998b738 1129extern int open_dev_flags(int devnum, int flags);
a931db9e 1130extern int open_dev_excl(int devnum);
8d80900b 1131extern int is_standard(char *dev, int *nump);
9008ed1c 1132extern int same_dev(char *one, char *two);
dd0781e5 1133
f1ae21c4 1134extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1135extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1136extern struct mddev_dev *conf_get_devs(void);
8382f19b 1137extern int conf_test_dev(char *devname);
4e8d9f0a 1138extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1139extern struct createinfo *conf_get_create_info(void);
1140extern void set_conffile(char *file);
1141extern char *conf_get_mailaddr(void);
1142extern char *conf_get_mailfrom(void);
1143extern char *conf_get_program(void);
0ac91628 1144extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1145extern char *conf_line(FILE *file);
98c6faba 1146extern char *conf_word(FILE *file, int allow_key);
0ac91628 1147extern int conf_name_is_free(char *name);
7c336758 1148extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1149extern int devname_matches(char *name, char *match);
2244d1a9
N
1150extern struct mddev_ident *conf_match(struct supertype *st,
1151 struct mdinfo *info,
1152 char *devname,
1153 int verbose, int *rvp);
6d11ec6f 1154extern int experimental(void);
0ac91628 1155
e0d19036
NB
1156extern void free_line(char *line);
1157extern int match_oneof(char *devices, char *devname);
e0d19036 1158extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1159extern const int uuid_zero[4];
f277ce36 1160extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1161extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1162extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1163extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1164 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1165extern unsigned long calc_csum(void *super, int bytes);
583315d9 1166extern int enough(int level, int raid_disks, int layout, int clean,
265e0f17 1167 char *avail, int avail_disks);
8453e704 1168extern int enough_fd(int fd);
e0d19036 1169extern int ask(char *mesg);
8fac0577 1170extern unsigned long long get_component_size(int fd);
0430ed48 1171extern void remove_partitions(int fd);
691c6ee1 1172extern int test_partition(int fd);
bfd76b93 1173extern int test_partition_from_id(dev_t id);
577e8448 1174extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1175extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1176 int chunksize, unsigned long long devsize);
edd8d13c
NB
1177extern int flush_metadata_updates(struct supertype *st);
1178extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c
DW
1179extern int assemble_container_content(struct supertype *st, int mdfd,
1180 struct mdinfo *content, int runstop,
49680258 1181 char *chosen_name, int verbose,
b76b30e0 1182 char *backup_file, int freeze_reshape);
326727d9
AC
1183extern struct mdinfo *container_choose_spares(struct supertype *st,
1184 unsigned long long min_size,
1185 struct domainlist *domlist,
1186 char *spare_group,
1187 const char *metadata, int get_one);
d52bb542 1188extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1189extern int add_disk(int mdfd, struct supertype *st,
1190 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1191extern int remove_disk(int mdfd, struct supertype *st,
1192 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1193extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1194unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1195
56eb10c0 1196extern char *human_size(long long bytes);
e4965ef8
N
1197extern char *human_size_brief(long long bytes);
1198extern void print_r10_layout(int layout);
e0d19036 1199
63152c1b
NB
1200#define NoMdDev (1<<23)
1201extern int find_free_devnum(int use_partitions);
1202
e0d19036
NB
1203extern void put_md_name(char *name);
1204extern char *get_md_name(int dev);
1205
c913b90e 1206extern char DefaultConfFile[];
b5e64645 1207
69207ff6
N
1208extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1209 char *chosen);
1210/* values for 'trustworthy' */
1211#define LOCAL 1
d1d3482b 1212#define LOCAL_ANY 10
69207ff6
N
1213#define FOREIGN 2
1214#define METADATA 3
6be1d39d 1215extern int open_mddev(char *dev, int report_errors);
a322f70c 1216extern int open_container(int fd);
33414a01
DW
1217extern int is_container_member(struct mdstat_ent *ent, char *devname);
1218extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1219extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1220extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1221
a931db9e 1222extern int mdmon_running(int devnum);
24f6f99b 1223extern int mdmon_pid(int devnum);
40ebbb9c 1224extern int check_env(char *name);
148acb7b 1225extern __u32 random32(void);
8850ee3e 1226extern int start_mdmon(int devnum);
a931db9e 1227
999b4972
N
1228extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1229 struct supertype *st, unsigned long stripes,
1230 int *fds, unsigned long long *offsets,
1231 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1232void abort_reshape(struct mdinfo *sra);
999b4972 1233
2f6079dc 1234extern char *devnum2devname(int num);
11877f4d 1235extern void fmt_devname(char *name, int num);
77472ff8 1236extern int devname2devnum(char *name);
c94709e8 1237extern int stat2devnum(struct stat *st);
2f6079dc 1238extern int fd2devnum(int fd);
b5e64645 1239
a931db9e
NB
1240static inline int dev2major(int d)
1241{
1242 if (d >= 0)
1243 return MD_MAJOR;
1244 else
1245 return get_mdp_major();
1246}
1247
1248static inline int dev2minor(int d)
1249{
1250 if (d >= 0)
1251 return d;
1252 return (-1-d) << MdpMinorShift;
1253}
1254
6416d527
NB
1255static inline int ROUND_UP(int a, int base)
1256{
1257 return ((a+base-1)/base)*base;
1258}
1259
3c558363
N
1260static inline int is_subarray(char *vers)
1261{
e9dd1598
N
1262 /* The version string for a 'subarray' (an array in a container)
1263 * is
1264 * /containername/componentname for normal read-write arrays
746a6567
N
1265 * -containername/componentname for arrays which mdmon must not
1266 * reconfigure. They might be read-only
1267 * or might be undergoing reshape etc.
e9dd1598
N
1268 * containername is e.g. md0, md_d1
1269 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1270 */
1271 return (*vers == '/' || *vers == '-');
3c558363
N
1272}
1273
e5408a32
DW
1274static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1275{
1276 return &ent->metadata_version[10+strlen(container)+1];
1277}
1278
2a24d7b6
DW
1279#ifdef DEBUG
1280#define dprintf(fmt, arg...) \
1281 fprintf(stderr, fmt, ##arg)
1282#else
1283#define dprintf(fmt, arg...) \
1284 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1285#endif
1a0ee0ba
DK
1286#include <assert.h>
1287#include <stdarg.h>
1288static inline int xasprintf(char **strp, const char *fmt, ...) {
1289 va_list ap;
1290 int ret;
1291 va_start(ap, fmt);
089485cb 1292 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1293 va_end(ap);
1294 assert(ret >= 0);
1295 return ret;
1296}
2a24d7b6 1297
b5e64645
NB
1298#define LEVEL_MULTIPATH (-4)
1299#define LEVEL_LINEAR (-1)
1300#define LEVEL_FAULTY (-5)
1301
17f25ca6
NB
1302/* kernel module doesn't know about these */
1303#define LEVEL_CONTAINER (-100)
a322f70c 1304#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1305
b5e64645
NB
1306
1307/* faulty stuff */
1308
1309#define WriteTransient 0
1310#define ReadTransient 1
1311#define WritePersistent 2
1312#define ReadPersistent 3
1313#define WriteAll 4 /* doesn't go to device */
1314#define ReadFixable 5
1315#define Modes 6
1316
1317#define ClearErrors 31
1318#define ClearFaults 30
1319
1320#define AllPersist 100 /* internal use only */
1321#define NoPersist 101
1322
1323#define ModeMask 0x1f
1324#define ModeShift 5
fe6729fa
NB
1325
1326
1327#ifdef __TINYC__
1328#undef minor
1329#undef major
1330#undef makedev
1331#define minor(x) ((x)&0xff)
1332#define major(x) (((x)>>8)&0xff)
1333#define makedev(M,m) (((M)<<8) | (m))
1334#endif
1335
b640a252 1336/* for raid4/5/6 */
e86c9dd6
NB
1337#define ALGORITHM_LEFT_ASYMMETRIC 0
1338#define ALGORITHM_RIGHT_ASYMMETRIC 1
1339#define ALGORITHM_LEFT_SYMMETRIC 2
1340#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1341
1342/* Define non-rotating (raid4) algorithms. These allow
1343 * conversion of raid4 to raid5.
1344 */
1345#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1346#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1347
1348/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1349 * Firstly, the exact positioning of the parity block is slightly
1350 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1351 * of DDF.
1352 * Secondly, or order of datablocks over which the Q syndrome is computed
1353 * is different.
1354 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1355 * These layouts are from the DDFv1.2 spec.
1356 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1357 * leaves RLQ=3 as 'Vendor Specific'
1358 */
1359
1360#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1361#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1362#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1363
1364
1365/* For every RAID5 algorithm we define a RAID6 algorithm
1366 * with exactly the same layout for data and parity, and
1367 * with the Q block always on the last device (N-1).
1368 * This allows trivial conversion from RAID5 to RAID6
1369 */
1370#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1371#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1372#define ALGORITHM_LEFT_SYMMETRIC_6 18
1373#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1374#define ALGORITHM_PARITY_0_6 20
1375#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1376
33a6535d
AW
1377/* Define PATH_MAX in case we don't use glibc or standard library does
1378 * not have PATH_MAX defined. Assume max path length is 4K characters.
1379 */
1380#ifndef PATH_MAX
1381#define PATH_MAX 4096
1382#endif
1383
2d3603ba
KW
1384#define PROCESS_DELAYED -2
1385#define PROCESS_PENDING -3