]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Avoid stack overflow if GPT partition entries on disk are > 128 bytes
[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 */
353632d9
NB
197 int reshape_active;
198 unsigned long long reshape_progress;
6e75048b
AK
199 int recovery_blocked; /* for external metadata it
200 * indicates that there is
201 * reshape in progress in
202 * container,
203 * for native metadata it is
204 * reshape_active field mirror
205 */
e1516be1
DW
206 union {
207 unsigned long long resync_start; /* per-array resync position */
208 unsigned long long recovery_start; /* per-device rebuild position */
b7528a20 209 #define MaxSector (~0ULL) /* resync/recovery complete position */
e1516be1 210 };
a67dd8cc 211 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 212 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 213 int errors;
f21e18ca 214 unsigned long cache_size; /* size of raid456 stripe cache*/
7e0f6979
NB
215 int mismatch_cnt;
216 char text_version[50];
6e46bf34
DW
217 void *update_private; /* for passing metadata-format
218 * specific update data
219 * between successive calls to
220 * update_super()
221 */
dd15dc4a
NB
222
223 int container_member; /* for assembling external-metatdata arrays
224 * This is to be used internally by metadata
225 * handler only */
97b4d0e9
DW
226 int container_enough; /* flag external handlers can set to
227 * indicate that subarrays have not enough (-1),
228 * enough to start (0), or all expected disks (1) */
06c7f68e 229 char sys_name[20];
7e0f6979 230 struct mdinfo *devs;
06c7f68e 231 struct mdinfo *next;
549e9569
NB
232
233 /* Device info for mdmon: */
e1516be1 234 int recovery_fd;
549e9569 235 int state_fd;
8d45d196
DW
236 #define DS_FAULTY 1
237 #define DS_INSYNC 2
238 #define DS_WRITE_MOSTLY 4
239 #define DS_SPARE 8
240 #define DS_BLOCKED 16
241 #define DS_REMOVE 1024
92967543 242 #define DS_UNBLOCK 2048
549e9569
NB
243 int prev_state, curr_state, next_state;
244
4b1ac34b
NB
245};
246
5bbb4842
NB
247struct createinfo {
248 int uid;
249 int gid;
250 int autof;
251 int mode;
38098016 252 int symlinks;
058574b1 253 struct supertype *supertype;
5bbb4842
NB
254};
255
9a9dab36 256#define Name "mdadm"
682c7051 257
e0d19036
NB
258enum mode {
259 ASSEMBLE=1,
260 BUILD,
261 CREATE,
262 MANAGE,
263 MISC,
264 MONITOR,
dd0781e5 265 GROW,
8382f19b 266 INCREMENTAL,
1f48664b 267 AUTODETECT,
e0d19036
NB
268};
269
64c4757e 270extern char short_options[];
024768c4 271extern char short_bitmap_options[];
c06487ce 272extern char short_bitmap_auto_options[];
64c4757e 273extern struct option long_options[];
56eedc1a 274extern char Version[], Usage[], Help[], OptionHelp[],
dd0781e5 275 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 276 Help_incr[],
e0d19036 277 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 278
997aed5d 279/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 280 * numbers later than any 'short' character option.
997aed5d
NB
281 */
282enum special_options {
1c7a808c 283 AssumeClean = 300,
997aed5d
NB
284 BitmapChunk,
285 WriteBehind,
286 ReAdd,
287 NoDegraded,
288 Sparc22,
1c7a808c 289 BackupFile,
997aed5d 290 HomeHost,
589395d6 291 AutoHomeHost,
38098016 292 Symlinks,
1f48664b 293 AutoDetect,
1770662b 294 Waitclean,
4cce4069 295 DetailPlatform,
33414a01 296 KillSubarray,
1c7a808c 297 UpdateSubarray,
edde9560 298 IncrementalPath,
1c7a808c
N
299 NoSharing,
300 HelpOptions,
301 Brief,
302 ManageOpt,
303 Add,
304 Remove,
305 Fail,
306 MiscOpt,
307 WaitOpt,
308 ConfigFile,
309 ChunkSize,
310 WriteMostly,
311 Layout,
312 Auto,
313 Force,
314 SuperMinor,
315 EMail,
316 ProgramOpt,
317 Increment,
318 Fork,
319 Bitmap,
320 RebuildMapOpt,
87f26d14 321 InvalidBackup,
20b60dcd 322 UdevRules,
b76b30e0 323 FreezeReshape,
2dddadb0 324 Continue,
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),
453 GET_DEVS = (1 << 10), /* gets role, major, minor */
454 GET_OFFSET = (1 << 11),
455 GET_SIZE = (1 << 12),
456 GET_STATE = (1 << 13),
457 GET_ERROR = (1 << 14),
dab4a513 458};
e86c9dd6
NB
459
460/* If fd >= 0, get the array it is open on,
461 * else use devnum. >=0 -> major9. <0.....
462 */
549e9569 463extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 464extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
465extern void sysfs_free(struct mdinfo *sra);
466extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
467extern int sysfs_attr_match(const char *attr, const char *str);
468extern int sysfs_match_word(const char *word, char **list);
7e0f6979 469extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 470 char *name, char *val);
7e0f6979 471extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 472 char *name, unsigned long long val);
97590376 473extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
474extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
475 char *name);
476extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 477extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 478 char *name, unsigned long long *val);
7236ee7a 479extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
480extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
481 char *name);
93ecfa01
N
482extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
483 char *name, char *val, int size);
8ed3e5e1 484extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 485extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 486extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 487extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 488extern int sysfs_unique_holder(int devnum, long rdev);
bc77ed53 489extern int sysfs_freeze_array(struct mdinfo *sra);
c69b251b 490extern int load_sys(char *path, char *buf);
130994cb
AK
491extern int reshape_prepare_fdlist(char *devname,
492 struct mdinfo *sra,
493 int raid_disks,
494 int nrdisks,
495 unsigned long blocks,
496 char *backup_file,
497 int *fdlist,
498 unsigned long long *offsets);
499extern void reshape_free_fdlist(int *fdlist,
500 unsigned long long *offsets,
501 int size);
e6e9d47b
AK
502extern int reshape_open_backup_file(char *backup,
503 int fd,
504 char *devname,
505 long blocks,
506 int *fdlist,
a93f87ee
N
507 unsigned long long *offsets,
508 int restart);
1c009fc2
AK
509extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
510 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
511
512extern int save_stripes(int *source, unsigned long long *offsets,
513 int raid_disks, int chunk_size, int level, int layout,
514 int nwrites, int *dest,
a6288483
N
515 unsigned long long start, unsigned long long length,
516 char *buf);
353632d9
NB
517extern int restore_stripes(int *dest, unsigned long long *offsets,
518 int raid_disks, int chunk_size, int level, int layout,
519 int source, unsigned long long read_offset,
2fcb75ae
AK
520 unsigned long long start, unsigned long long length,
521 char *src_buf);
e86c9dd6 522
52826846
NB
523#ifndef Sendmail
524#define Sendmail "/usr/lib/sendmail -t"
525#endif
526
773135f5
NB
527#define SYSLOG_FACILITY LOG_DAEMON
528
682c7051
NB
529extern char *map_num(mapping_t *map, int num);
530extern int map_name(mapping_t *map, char *name);
b640a252 531extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 532
16c6fa80 533extern char *map_dev(int major, int minor, int create);
64c4757e 534
549e9569 535struct active_array;
2e735d19 536struct metadata_update;
64c4757e 537
999b4972
N
538
539/* 'struct reshape' records the intermediate states
540 * a general reshape.
541 * The starting geometry is converted to the 'before' geometry
542 * by at most an atomic level change. They could be the same.
543 * Similarly the 'after' geometry is converted to the final
544 * geometry by at most a level change.
545 * Note that 'before' and 'after' must have the same level.
546 * 'blocks' is the minimum number of sectors for a reshape unit.
547 * This will be a multiple of the stripe size in each of the
548 * 'before' and 'after' geometries.
549 * If 'blocks' is 0, no restriping is necessary.
550 */
551struct reshape {
552 int level;
553 int parity; /* number of parity blocks/devices */
554 struct {
555 int layout;
556 int data_disks;
557 } before, after;
558 unsigned long long backup_blocks;
559 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
560 unsigned long long new_size; /* New size of array in sectors */
561};
562
6adfd3af
NB
563/* A superswitch provides entry point the a metadata handler.
564 *
565 * The super_switch primarily operates on some "metadata" that
566 * is accessed via the 'supertype'.
567 * This metadata has one of three possible sources.
568 * 1/ It is read from a single device. In this case it may not completely
569 * describe the array or arrays as some information might be on other
570 * devices.
571 * 2/ It is read from all devices in a container. In this case all
572 * information is present.
573 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
574 * be complete once enough ->add_to_super calls have completed.
575 *
576 * When creating an array inside a container, the metadata will be
577 * formed by a combination of 2 and 3. The metadata or the array is read,
578 * then new information is added.
579 *
580 * The metadata must sometimes have a concept of a 'current' array
581 * and a 'current' device.
582 * The 'current' array is set by init_super to be the newly created array,
583 * or is set by super_by_fd when it finds it is looking at an array inside
584 * a container.
585 *
586 * The 'current' device is either the device that the metadata was read from
587 * in case 1, or the last device added by add_to_super in case 3.
588 * Case 2 does not identify a 'current' device.
589 */
f9ce90ba 590extern struct superswitch {
6adfd3af
NB
591
592 /* Used to report details of metadata read from a component
593 * device. ->load_super has been called.
594 */
3da92f27 595 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 596 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 597 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 598 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
599
600 /* Used to report details of an active array.
601 * ->load_super was possibly given a 'component' string.
602 */
3da92f27 603 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 604 void (*brief_detail_super)(struct supertype *st);
0d726f17 605 void (*export_detail_super)(struct supertype *st);
6adfd3af 606
4cce4069 607 /* Optional: platform hardware / firmware details */
5615172f 608 int (*detail_platform)(int verbose, int enumerate_only);
4cce4069 609
6adfd3af
NB
610 /* Used:
611 * to get uuid to storing in bitmap metadata
612 * and 'reshape' backup-data metadata
613 * To see if a device is being re-added to an array it was part of.
614 */
3da92f27 615 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 616
8592f29d 617 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
618 * the container, or about an individual array within the container.
619 * The determination is made either by:
620 * load_super being given a 'component' string.
621 * validate_geometry determining what to create.
d2ca6449
NB
622 * The info includes both array information and device information.
623 * The particular device should be:
624 * The last device added by add_to_super
625 * The device the metadata was loaded from by load_super
a5d85af7
N
626 * If 'map' is present, then it is an array raid_disks long
627 * (raid_disk must already be set and correct) and it is filled
628 * with 1 for slots that are thought to be active and 0 for slots which
629 * appear to be failed/missing.
95eeceeb 630 * *info is zeroed out before data is added.
6adfd3af 631 */
a5d85af7 632 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 633 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 634 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 635 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 636 */
3da92f27 637 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
638
639 /* Make one of several generic modifications to metadata
640 * prior to assembly (or other times).
641 * sparc2.2 - first bug in early 0.90 metadata
642 * super-minor - change name of 0.90 metadata
643 * summaries - 'correct' any redundant data
644 * resync - mark array as dirty to trigger a resync.
645 * uuid - set new uuid - only 0.90 or 1.x
646 * name - change the name of the array (where supported)
647 * homehost - change which host this array is tied to.
648 * devicesize - If metadata is at start of device, change recorded
649 * device size to match actual device size
650 * byteorder - swap bytes for 0.90 metadata
651 *
652 * force-one - mark that device as uptodate, not old or failed.
653 * force-array - mark array as clean if it would not otherwise
654 * assemble
655 * assemble - not sure how this is different from force-one...
656 * linear-grow-new - add a new device to a linear array, but don't
657 * change the size: so superblock still matches
658 * linear-grow-update - now change the size of the array.
16715c01
DL
659 * writemostly - set the WriteMostly1 bit in the superblock devflags
660 * readwrite - clear the WriteMostly1 bit in the superblock devflags
6adfd3af 661 */
68c7d6d7 662 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 663 char *update,
e5eac01f
NB
664 char *devname, int verbose,
665 int uuid_set, char *homehost);
6adfd3af
NB
666
667 /* Create new metadata for new array as described. This could
668 * be a new container, or an array in a pre-existing container.
669 * Also used to zero metadata prior to writing it to invalidate old
670 * metadata.
671 */
3da92f27
NB
672 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
673 unsigned long long size, char *name,
674 char *homehost, int *uuid);
6adfd3af
NB
675
676 /* update the metadata to include new device, either at create or
677 * when hot-adding a spare.
678 */
f20c3968 679 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
1a64be56
LM
680 int fd, char *devname);
681 /* update the metadata to delete a device,
682 * when hot-removing.
683 */
684 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
685
686 /* Write metadata to one device when fixing problems or adding
687 * a new device.
688 */
3da92f27 689 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
690
691 /* Write all metadata for this array.
692 */
111d01fc 693 int (*write_init_super)(struct supertype *st);
64557c33 694 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 695 int (*load_super)(struct supertype *st, int fd, char *devname);
2b959fbf 696 int (*load_container)(struct supertype *st, int fd, char *devname);
82d9eba6 697 struct supertype * (*match_metadata_desc)(char *arg);
1bf4e2d9 698 __u64 (*avail_size)(struct supertype *st, __u64 size);
80e7f8c3 699 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
3da92f27 700 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 701 int delay, int write_behind,
f9c25f1d 702 unsigned long long size, int may_change, int major);
3da92f27
NB
703 void (*locate_bitmap)(struct supertype *st, int fd);
704 int (*write_bitmap)(struct supertype *st, int fd);
705 void (*free_super)(struct supertype *st);
78e44928
NB
706
707 /* validate_geometry is called with an st returned by
708 * match_metadata_desc.
709 * It should check that the geometry described in compatible with
710 * the metadata type. It will be called repeatedly as devices
711 * added to validate changing size and new devices. If there are
712 * inter-device dependencies, it should record sufficient details
713 * so these can be validated.
30f58b22 714 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
715 * Return value is:
716 * 1: everything is OK
717 * 0: not OK for some reason - if 'verbose', then error was reported.
718 * -1: st->sb was NULL, 'subdev' is a member of a container of this
719 * types, but array is not acceptable for some reason
720 * message was reported even if verbose is 0.
78e44928 721 */
17f25ca6
NB
722 int (*validate_geometry)(struct supertype *st, int level, int layout,
723 int raiddisks,
c21e737b 724 int *chunk, unsigned long long size,
2c514b71
NB
725 char *subdev, unsigned long long *freesize,
726 int verbose);
598f0d58 727
00bbdbda 728 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
729 /* query the supertype for default geometry */
730 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
731 /* Permit subarray's to be deleted from inactive containers */
732 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 733 /* Permit subarray's to be modified */
a951a4f7 734 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 735 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
736 /* Check if reshape is supported for this external format.
737 * st is obtained from super_by_fd() where st->subarray[0] is
738 * initialized to indicate if reshape is being performed at the
739 * container or subarray level
740 */
741 int (*reshape_super)(struct supertype *st, long long size, int level,
742 int layout, int chunksize, int raid_disks,
41784c88
AK
743 int delta_disks, char *backup, char *dev,
744 int verbose); /* optional */
999b4972
N
745 int (*manage_reshape)( /* optional */
746 int afd, struct mdinfo *sra, struct reshape *reshape,
747 struct supertype *st, unsigned long blocks,
748 int *fds, unsigned long long *offsets,
749 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 750
549e9569 751/* for mdmon */
cba0191b
NB
752 int (*open_new)(struct supertype *c, struct active_array *a,
753 char *inst);
ed9d66aa
NB
754
755 /* Tell the metadata handler the current state of the array.
756 * This covers whether it is known to be consistent (no pending writes)
01f157d7 757 * and how far along a resync is known to have progressed
ed9d66aa
NB
758 * (in a->resync_start).
759 * resync status is really irrelevant if the array is not consistent,
760 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
761 * If 'consistent' is '2', then the array can mark it dirty if a
762 * resync/recovery/whatever is required, or leave it clean if not.
763 * Return value is 0 dirty (not consistent) and 1 if clean.
764 * it is only really important if consistent is passed in as '2'.
4e5528c6 765 */
01f157d7 766 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
767
768 /* When the state of a device might have changed, we call set_disk to
769 * tell the metadata what the current state is.
770 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
771 * transitions.
772 * set_disk might be called when the state of the particular disk has
773 * not in fact changed.
774 */
8d45d196 775 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
776 void (*sync_metadata)(struct supertype *st);
777 void (*process_update)(struct supertype *st,
778 struct metadata_update *update);
edd8d13c
NB
779 void (*prepare_update)(struct supertype *st,
780 struct metadata_update *update);
549e9569 781
6c3fb95c
NB
782 /* activate_spare will check if the array is degraded and, if it
783 * is, try to find some spare space in the container.
784 * On success, it add appropriate updates (For process_update) to
785 * to the 'updates' list and returns a list of 'mdinfo' identifying
786 * the device, or devices as there might be multiple missing
787 * devices and multiple spares available.
788 */
789 struct mdinfo *(*activate_spare)(struct active_array *a,
790 struct metadata_update **updates);
2cda7640
ML
791 /*
792 * Return statically allocated string that represents metadata specific
793 * controller domain of the disk. The domain is used in disk domain
794 * matching functions. Disks belong to the same domain if the they have
795 * the same domain from mdadm.conf and belong the same metadata domain.
796 * Returning NULL or not providing this handler means that metadata
797 * does not distinguish the differences between disks that belong to
798 * different controllers. They are in the domain specified by
799 * configuration file (mdadm.conf).
800 * In case when the metadata has the notion of domains based on disk
801 * it shall return NULL for disks that do not belong to the controller
802 * the supported domains. Such disks will form another domain and won't
803 * be mixed with supported ones.
804 */
805 const char *(*get_disk_controller_domain)(const char *path);
549e9569 806
ba53ea59
AK
807 /* for external backup area */
808 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
809
f277ce36 810 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 811 int external;
4cce4069 812 const char *name; /* canonical metadata name */
0f22b998 813} *superlist[];
82d9eba6 814
0f22b998
N
815extern struct superswitch super0, super1;
816extern struct superswitch super_imsm, super_ddf;
0592faeb 817extern struct superswitch mbr, gpt;
cdddbdbc 818
edd8d13c
NB
819struct metadata_update {
820 int len;
821 char *buf;
822 void *space; /* allocated space that monitor will use */
cb23f1f4
N
823 void **space_list; /* list of allocated spaces that monitor can
824 * use or that it returned.
825 */
edd8d13c
NB
826 struct metadata_update *next;
827};
828
6adfd3af
NB
829/* A supertype holds a particular collection of metadata.
830 * It identifies the metadata type by the superswitch, and the particular
831 * sub-version of that metadata type.
832 * metadata read in or created is stored in 'sb' and 'info'.
833 * There are also fields used by mdmon to track containers.
834 *
352452c3
N
835 * A supertype may refer to:
836 * Just an array, possibly in a container
837 * A container, not identifying any particular array
838 * Info read from just one device, not yet fully describing the array/container.
839 *
840 *
6adfd3af
NB
841 * A supertype is created by:
842 * super_by_fd
843 * guess_super
844 * dup_super
845 */
82d9eba6
NB
846struct supertype {
847 struct superswitch *ss;
848 int minor_version;
ea329559 849 int max_devs;
2f6079dc 850 int container_dev; /* devnum of container */
64557c33 851 void *sb;
111d01fc 852 void *info;
df3346e6
LM
853 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
854 HW/firmware related incompatability to load metadata.
855 Used when examining metadata to display content of disk
856 when user has no hw/firmare compatible system.
857 */
edd8d13c
NB
858 struct metadata_update *updates;
859 struct metadata_update **update_tail;
860
549e9569
NB
861 /* extra stuff used by mdmon */
862 struct active_array *arrays;
549e9569 863 int sock; /* listen to external programs */
549e9569
NB
864 int devnum;
865 char *devname; /* e.g. md0. This appears in metadata_verison:
866 * external:/md0/12
867 */
868 int devcnt;
869
870 struct mdinfo *devs;
871
82d9eba6 872};
f9ce90ba 873
4725bc31 874extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
875enum guess_types { guess_any, guess_array, guess_partitions };
876extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
877static inline struct supertype *guess_super(int fd) {
878 return guess_super_type(fd, guess_any);
879}
3da92f27 880extern struct supertype *dup_super(struct supertype *st);
beae1dfe 881extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 882extern int must_be_container(int fd);
0f0749ad 883extern int dev_size_from_id(dev_t id, unsigned long long *size);
8382f19b
NB
884extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
885 mdu_disk_info_t *disk);
a7c6e3fb 886void wait_for(char *dev, int fd);
f9ce90ba 887
5527fc74
N
888/*
889 * Data structures for policy management.
890 * Each device can have a policy structure that lists
891 * various name/value pairs each possibly with a metadata associated.
892 * The policy list is sorted by name/value/metadata
893 */
894struct dev_policy {
895 struct dev_policy *next;
896 char *name; /* None of these strings are allocated. They are
897 * all just references to strings which are known
898 * to exist elsewhere.
899 * name and metadata can be compared by address equality.
900 */
901 const char *metadata;
2cda7640 902 const char *value;
5527fc74
N
903};
904
4e8d9f0a 905extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
906
907/* iterate over the sublist starting at list, having the same
908 * 'name' as 'list', and matching the given metadata (Where
909 * NULL matches anything
910 */
911#define pol_for_each(item, list, _metadata) \
912 for (item = list; \
913 item && item->name == list->name; \
914 item = item->next) \
915 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
916 ; else
917
918/*
919 * policy records read from mdadm are largely just name-value pairs.
920 * The names are constants, not strdupped
921 */
922struct pol_rule {
923 struct pol_rule *next;
924 char *type; /* rule_policy or rule_part */
925 struct rule {
926 struct rule *next;
927 char *name;
928 char *value;
929 char *dups; /* duplicates of 'value' with a partNN appended */
930 } *rule;
931};
932
933extern char rule_policy[], rule_part[];
934extern char rule_path[], rule_type[];
935extern char type_part[], type_disk[];
936
937extern void policyline(char *line, char *type);
4e8d9f0a 938extern void policy_add(char *type, ...);
5527fc74
N
939extern void policy_free(void);
940
73c9c47c 941extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 942extern struct dev_policy *disk_policy(struct mdinfo *disk);
4e8d9f0a 943extern struct dev_policy *devnum_policy(int dev);
5527fc74
N
944extern void dev_policy_free(struct dev_policy *p);
945
e78dda3b
N
946//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
947extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
948extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
949
e3bb5f14
N
950enum policy_action {
951 act_default,
952 act_include,
953 act_re_add,
d2db3045
N
954 act_spare, /* This only applies to bare devices */
955 act_spare_same_slot, /* this allows non-bare devices,
956 * but only if recent removal */
957 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
958 act_err
959};
960
961extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
962 enum policy_action want);
963extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
964 enum policy_action want);
f5f12c84
N
965
966struct domainlist {
967 struct domainlist *next;
2cda7640 968 const char *dom;
f5f12c84
N
969};
970
971extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
972 const char *metadata);
973extern struct domainlist *domain_from_array(struct mdinfo *mdi,
974 const char *metadata);
e78dda3b
N
975extern void domainlist_add_dev(struct domainlist **dom, int devnum,
976 const char *metadata);
f5f12c84
N
977extern void domain_free(struct domainlist *dl);
978extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
979 const char *metadata);
e78dda3b 980void domain_add(struct domainlist **domp, char *domain);
f5f12c84 981
403410eb
PC
982extern void policy_save_path(char *id_path, struct map_ent *array);
983extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
984
838acbc2
NB
985#if __GNUC__ < 3
986struct stat64;
987#endif
988
45e878bb
NB
989#define HAVE_NFTW we assume
990#define HAVE_FTW
991
2df1f269 992#ifdef __UCLIBC__
45e878bb 993# include <features.h>
2df1f269
BRF
994# ifndef __UCLIBC_HAS_LFS__
995# define lseek64 lseek
996# endif
45e878bb
NB
997# ifndef __UCLIBC_HAS_FTW__
998# undef HAVE_FTW
999# undef HAVE_NFTW
1000# endif
1001#endif
1002
1003#ifdef __dietlibc__
1004# undef HAVE_NFTW
1005#endif
1006
e39b673e 1007#if defined(__KLIBC__)
1008# undef HAVE_NFTW
1009# undef HAVE_FTW
1010#endif
1011
45e878bb 1012#ifndef HAVE_NFTW
173fc515 1013# define FTW_PHYS 1
45e878bb
NB
1014# ifndef HAVE_FTW
1015 struct FTW {};
173fc515
NB
1016# endif
1017#endif
1018
45e878bb
NB
1019#ifdef HAVE_FTW
1020# include <ftw.h>
1021#endif
1022
173fc515
NB
1023extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1024
f9ce90ba 1025
64c4757e 1026extern int Manage_ro(char *devname, int fd, int readonly);
91f068bf 1027extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
dd0781e5 1028extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
64c4757e 1029extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1030 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1031 char *update, int force);
1f48664b 1032extern int autodetect(void);
e5329c37 1033extern int Grow_Add_device(char *devname, int fd, char *newdev);
8fac0577 1034extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
06b0d786 1035extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
e86c9dd6 1036 long long size,
691a36b7 1037 int level, char *layout_str, int chunksize, int raid_disks,
e2e53a2d 1038 struct mddev_dev *devlist,
ce52f92f 1039 int assume_clean, int force);
06b0d786 1040extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1041 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1042extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1043 struct mdinfo *info, char *backup_file,
1044 int freeze_reshape);
1045
3f54bd62
AK
1046extern int restore_backup(struct supertype *st,
1047 struct mdinfo *content,
1048 int working_disks,
1049 int spares,
1050 char *backup_file,
1051 int verbose);
2dddadb0
AK
1052extern int Grow_continue_command(char *devname, int fd,
1053 char *backup_file, int verbose);
64c4757e 1054
7f91af49 1055extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1056 struct mddev_ident *ident,
87f26d14
N
1057 struct mddev_dev *devlist,
1058 char *backup_file, int invalid_backup,
64c4757e 1059 int readonly, int runstop,
0ac91628 1060 char *update, char *homehost, int require_homehost,
b76b30e0 1061 int verbose, int force, int freeze_reshape);
64c4757e 1062
7f91af49 1063extern int Build(char *mddev, int chunk, int level, int layout,
a655e550 1064 int raiddisks, struct mddev_dev *devlist, int assume_clean,
7f91af49 1065 char *bitmap_file, int bitmap_chunk, int write_behind,
83208785 1066 int delay, int verbose, int autof, unsigned long long size);
64c4757e
NB
1067
1068
7f91af49 1069extern int Create(struct supertype *st, char *mddev,
5dd497ee 1070 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
3d3dd91e 1071 char *name, char *homehost, int *uuid,
a655e550 1072 int subdevs, struct mddev_dev *devlist,
47d79ef8 1073 int runstop, int verbose, int force, int assume_clean,
7f91af49 1074 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
64c4757e 1075
54bad364 1076extern int Detail(char *dev, int brief, int export, int test, char *homehost);
4cce4069 1077extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
e0d19036 1078extern int Query(char *dev);
a655e550 1079extern int Examine(struct mddev_dev *devlist, int brief, int export, int scan,
0d726f17 1080 int SparcAdjust, struct supertype *forcest, char *homehost);
a655e550 1081extern int Monitor(struct mddev_dev *devlist,
e0d19036 1082 char *mailaddr, char *alert_cmd,
aa88f531 1083 int period, int daemonise, int scan, int oneshot,
edde9560
AC
1084 int dosyslog, int test, char *pidfile, int increments,
1085 int share);
64c4757e 1086
9277cc77 1087extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
33414a01 1088extern int Kill_subarray(char *dev, char *subarray, int quiet);
fa56eddb 1089extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1090extern int Wait(char *dev);
9f1da824 1091extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1092
8382f19b 1093extern int Incremental(char *devname, int verbose, int runstop,
0ac91628 1094 struct supertype *st, char *homehost, int require_homehost,
b76b30e0 1095 int autof, int freeze_reshape);
8382f19b
NB
1096extern void RebuildMap(void);
1097extern int IncrementalScan(int verbose);
950bc344 1098extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1099extern int CreateBitmap(char *filename, int force, char uuid[16],
1100 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1101 unsigned long write_behind,
dcec9ee5
NB
1102 unsigned long long array_size,
1103 int major);
55935d51 1104extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1105extern int Write_rules(char *rule_name);
bf4fb153 1106extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1107extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 1108
64c4757e 1109extern int md_get_version(int fd);
e0d19036 1110extern int get_linux_version(void);
bc77ed53 1111extern int mdadm_version(char *version);
84e11361 1112extern long long parse_size(char *size);
64c4757e 1113extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1114extern int parse_layout_10(char *layout);
1115extern int parse_layout_faulty(char *layout);
682c7051
NB
1116extern int check_ext2(int fd, char *name);
1117extern int check_reiser(int fd, char *name);
1118extern int check_raid(int fd, char *name);
53ed6ac3
KW
1119extern int check_partitions(int fd, char *dname,
1120 unsigned long long freesize,
1121 unsigned long long size);
64c4757e 1122
dd0781e5 1123extern int get_mdp_major(void);
8b0dabea 1124extern int dev_open(char *dev, int flags);
e8a70c89 1125extern int open_dev(int devnum);
d998b738 1126extern int open_dev_flags(int devnum, int flags);
a931db9e 1127extern int open_dev_excl(int devnum);
8d80900b 1128extern int is_standard(char *dev, int *nump);
9008ed1c 1129extern int same_dev(char *one, char *two);
dd0781e5 1130
f1ae21c4 1131extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1132extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1133extern struct mddev_dev *conf_get_devs(void);
8382f19b 1134extern int conf_test_dev(char *devname);
4e8d9f0a 1135extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1136extern struct createinfo *conf_get_create_info(void);
1137extern void set_conffile(char *file);
1138extern char *conf_get_mailaddr(void);
1139extern char *conf_get_mailfrom(void);
1140extern char *conf_get_program(void);
0ac91628 1141extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1142extern char *conf_line(FILE *file);
98c6faba 1143extern char *conf_word(FILE *file, int allow_key);
0ac91628
N
1144extern int conf_name_is_free(char *name);
1145extern int devname_matches(char *name, char *match);
fa56eddb 1146extern struct mddev_ident *conf_match(struct mdinfo *info, struct supertype *st);
6d11ec6f 1147extern int experimental(void);
0ac91628 1148
e0d19036
NB
1149extern void free_line(char *line);
1150extern int match_oneof(char *devices, char *devname);
e0d19036 1151extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1152extern const int uuid_zero[4];
f277ce36 1153extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1154extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1155extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1156extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1157 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1158extern unsigned long calc_csum(void *super, int bytes);
583315d9 1159extern int enough(int level, int raid_disks, int layout, int clean,
265e0f17 1160 char *avail, int avail_disks);
8453e704 1161extern int enough_fd(int fd);
e0d19036 1162extern int ask(char *mesg);
8fac0577 1163extern unsigned long long get_component_size(int fd);
0430ed48 1164extern void remove_partitions(int fd);
691c6ee1 1165extern int test_partition(int fd);
bfd76b93 1166extern int test_partition_from_id(dev_t id);
577e8448 1167extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1168extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1169 int chunksize, unsigned long long devsize);
edd8d13c
NB
1170extern int flush_metadata_updates(struct supertype *st);
1171extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c
DW
1172extern int assemble_container_content(struct supertype *st, int mdfd,
1173 struct mdinfo *content, int runstop,
49680258 1174 char *chosen_name, int verbose,
b76b30e0 1175 char *backup_file, int freeze_reshape);
326727d9
AC
1176extern struct mdinfo *container_choose_spares(struct supertype *st,
1177 unsigned long long min_size,
1178 struct domainlist *domlist,
1179 char *spare_group,
1180 const char *metadata, int get_one);
d52bb542 1181extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1182extern int add_disk(int mdfd, struct supertype *st,
1183 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1184extern int remove_disk(int mdfd, struct supertype *st,
1185 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1186extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1187unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1188
56eb10c0 1189extern char *human_size(long long bytes);
e4965ef8
N
1190extern char *human_size_brief(long long bytes);
1191extern void print_r10_layout(int layout);
e0d19036 1192
63152c1b
NB
1193#define NoMdDev (1<<23)
1194extern int find_free_devnum(int use_partitions);
1195
e0d19036
NB
1196extern void put_md_name(char *name);
1197extern char *get_md_name(int dev);
1198
c913b90e 1199extern char DefaultConfFile[];
b5e64645 1200
69207ff6
N
1201extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1202 char *chosen);
1203/* values for 'trustworthy' */
1204#define LOCAL 1
d1d3482b 1205#define LOCAL_ANY 10
69207ff6
N
1206#define FOREIGN 2
1207#define METADATA 3
6be1d39d 1208extern int open_mddev(char *dev, int report_errors);
a322f70c 1209extern int open_container(int fd);
33414a01
DW
1210extern int is_container_member(struct mdstat_ent *ent, char *devname);
1211extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1212extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1213extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1214
a931db9e 1215extern int mdmon_running(int devnum);
24f6f99b 1216extern int mdmon_pid(int devnum);
40ebbb9c 1217extern int check_env(char *name);
148acb7b 1218extern __u32 random32(void);
8850ee3e 1219extern int start_mdmon(int devnum);
a931db9e 1220
999b4972
N
1221extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1222 struct supertype *st, unsigned long stripes,
1223 int *fds, unsigned long long *offsets,
1224 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1225void abort_reshape(struct mdinfo *sra);
999b4972 1226
2f6079dc 1227extern char *devnum2devname(int num);
11877f4d 1228extern void fmt_devname(char *name, int num);
77472ff8 1229extern int devname2devnum(char *name);
c94709e8 1230extern int stat2devnum(struct stat *st);
2f6079dc 1231extern int fd2devnum(int fd);
b5e64645 1232
a931db9e
NB
1233static inline int dev2major(int d)
1234{
1235 if (d >= 0)
1236 return MD_MAJOR;
1237 else
1238 return get_mdp_major();
1239}
1240
1241static inline int dev2minor(int d)
1242{
1243 if (d >= 0)
1244 return d;
1245 return (-1-d) << MdpMinorShift;
1246}
1247
6416d527
NB
1248static inline int ROUND_UP(int a, int base)
1249{
1250 return ((a+base-1)/base)*base;
1251}
1252
3c558363
N
1253static inline int is_subarray(char *vers)
1254{
e9dd1598
N
1255 /* The version string for a 'subarray' (an array in a container)
1256 * is
1257 * /containername/componentname for normal read-write arrays
746a6567
N
1258 * -containername/componentname for arrays which mdmon must not
1259 * reconfigure. They might be read-only
1260 * or might be undergoing reshape etc.
e9dd1598
N
1261 * containername is e.g. md0, md_d1
1262 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1263 */
1264 return (*vers == '/' || *vers == '-');
3c558363
N
1265}
1266
e5408a32
DW
1267static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1268{
1269 return &ent->metadata_version[10+strlen(container)+1];
1270}
1271
2a24d7b6
DW
1272#ifdef DEBUG
1273#define dprintf(fmt, arg...) \
1274 fprintf(stderr, fmt, ##arg)
1275#else
1276#define dprintf(fmt, arg...) \
1277 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1278#endif
1a0ee0ba
DK
1279#include <assert.h>
1280#include <stdarg.h>
1281static inline int xasprintf(char **strp, const char *fmt, ...) {
1282 va_list ap;
1283 int ret;
1284 va_start(ap, fmt);
089485cb 1285 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1286 va_end(ap);
1287 assert(ret >= 0);
1288 return ret;
1289}
2a24d7b6 1290
b5e64645
NB
1291#define LEVEL_MULTIPATH (-4)
1292#define LEVEL_LINEAR (-1)
1293#define LEVEL_FAULTY (-5)
1294
17f25ca6
NB
1295/* kernel module doesn't know about these */
1296#define LEVEL_CONTAINER (-100)
a322f70c 1297#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1298
b5e64645
NB
1299
1300/* faulty stuff */
1301
1302#define WriteTransient 0
1303#define ReadTransient 1
1304#define WritePersistent 2
1305#define ReadPersistent 3
1306#define WriteAll 4 /* doesn't go to device */
1307#define ReadFixable 5
1308#define Modes 6
1309
1310#define ClearErrors 31
1311#define ClearFaults 30
1312
1313#define AllPersist 100 /* internal use only */
1314#define NoPersist 101
1315
1316#define ModeMask 0x1f
1317#define ModeShift 5
fe6729fa
NB
1318
1319
1320#ifdef __TINYC__
1321#undef minor
1322#undef major
1323#undef makedev
1324#define minor(x) ((x)&0xff)
1325#define major(x) (((x)>>8)&0xff)
1326#define makedev(M,m) (((M)<<8) | (m))
1327#endif
1328
b640a252 1329/* for raid4/5/6 */
e86c9dd6
NB
1330#define ALGORITHM_LEFT_ASYMMETRIC 0
1331#define ALGORITHM_RIGHT_ASYMMETRIC 1
1332#define ALGORITHM_LEFT_SYMMETRIC 2
1333#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1334
1335/* Define non-rotating (raid4) algorithms. These allow
1336 * conversion of raid4 to raid5.
1337 */
1338#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1339#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1340
1341/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1342 * Firstly, the exact positioning of the parity block is slightly
1343 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1344 * of DDF.
1345 * Secondly, or order of datablocks over which the Q syndrome is computed
1346 * is different.
1347 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1348 * These layouts are from the DDFv1.2 spec.
1349 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1350 * leaves RLQ=3 as 'Vendor Specific'
1351 */
1352
1353#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1354#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1355#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1356
1357
1358/* For every RAID5 algorithm we define a RAID6 algorithm
1359 * with exactly the same layout for data and parity, and
1360 * with the Q block always on the last device (N-1).
1361 * This allows trivial conversion from RAID5 to RAID6
1362 */
1363#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1364#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1365#define ALGORITHM_LEFT_SYMMETRIC_6 18
1366#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1367#define ALGORITHM_PARITY_0_6 20
1368#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1369
33a6535d
AW
1370/* Define PATH_MAX in case we don't use glibc or standard library does
1371 * not have PATH_MAX defined. Assume max path length is 4K characters.
1372 */
1373#ifndef PATH_MAX
1374#define PATH_MAX 4096
1375#endif
1376
2d3603ba
KW
1377#define PROCESS_DELAYED -2
1378#define PROCESS_PENDING -3