]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Fix sign extension of bitmap_offset in super1.c
[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 };
fbdef498 214 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,
c2ecf5f6 325 Prefer,
997aed5d
NB
326};
327
64c4757e 328/* structures read from config file */
52826846
NB
329/* List of mddevice names and identifiers
330 * Identifiers can be:
331 * uuid=128-hex-uuid
332 * super-minor=decimal-minor-number-from-superblock
333 * devices=comma,separated,list,of,device,names,with,wildcards
334 *
335 * If multiple fields are present, the intersection of all matching
336 * devices is considered
337 */
98c6faba 338#define UnSet (0xfffe)
fa56eddb 339struct mddev_ident {
dd0781e5 340 char *devname;
aba69144 341
dd0781e5 342 int uuid_set;
3fa06e9d 343 int uuid[4];
947fd4dd 344 char name[33];
52826846 345
f21e18ca 346 int super_minor;
52826846 347
dd0781e5 348 char *devices; /* comma separated list of device
52826846
NB
349 * names with wild cards
350 */
dd0781e5 351 int level;
f21e18ca
N
352 int raid_disks;
353 int spare_disks;
82d9eba6 354 struct supertype *st;
dd0781e5
NB
355 int autof; /* 1 for normal, 2 for partitioned */
356 char *spare_group;
7ef02d01 357 char *bitmap_file;
c82f047c 358 int bitmap_fd;
dd0781e5 359
1771a6e2
N
360 char *container; /* /dev/whatever name of container, or
361 * uuid of container. You would expect
362 * this to be the 'devname' or UUID
dbb44303
N
363 * of some other entry.
364 */
365 char *member; /* subarray within a container */
366
fa56eddb 367 struct mddev_ident *next;
b179246f
N
368 union {
369 /* fields needed by different users of this structure */
370 int assembled; /* set when assembly succeeds */
371 };
fa56eddb 372};
64c4757e
NB
373
374/* List of device names - wildcards expanded */
a655e550 375struct mddev_dev {
64c4757e 376 char *devname;
1c7a808c 377 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
cd29a5c8
NB
378 * Not set for names read from .config
379 */
b3d31955 380 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
fe80f49b 381 char re_add;
811e6cbe 382 char used; /* set when used */
a655e550
N
383 struct mddev_dev *next;
384};
64c4757e 385
682c7051
NB
386typedef struct mapping {
387 char *name;
388 int num;
389} mapping_t;
390
e0d19036
NB
391
392struct mdstat_ent {
393 char *dev;
394 int devnum;
395 int active;
396 char *level;
397 char *pattern; /* U or up, _ for down */
398 int percent; /* -1 if no resync */
f94c116f 399 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
400 int devcnt;
401 int raid_disks;
549e9569 402 char * metadata_version;
3b57c466
N
403 struct dev_member {
404 char *name;
405 struct dev_member *next;
406 } *members;
e0d19036
NB
407 struct mdstat_ent *next;
408};
409
22a88995 410extern struct mdstat_ent *mdstat_read(int hold, int start);
e0d19036 411extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 412extern void mdstat_wait(int seconds);
58a4ba2a 413extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
8382f19b 414extern int mddev_busy(int devnum);
3b57c466 415extern struct mdstat_ent *mdstat_by_component(char *name);
78b10e66 416extern struct mdstat_ent *mdstat_by_subdev(char *subdev, int container);
8382f19b
NB
417
418struct map_ent {
419 struct map_ent *next;
420 int devnum;
1522c538 421 char metadata[20];
8382f19b 422 int uuid[4];
195254b8 423 int bad;
8382f19b
NB
424 char *path;
425};
1522c538 426extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
8382f19b 427 int uuid[4], char *path);
4eb26970 428extern void map_remove(struct map_ent **map, int devnum);
8382f19b 429extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
f2e55ecc
N
430extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
431extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
432extern void map_read(struct map_ent **melp);
433extern int map_write(struct map_ent *mel);
434extern void map_delete(struct map_ent **mapp, int devnum);
435extern void map_free(struct map_ent *map);
436extern void map_add(struct map_ent **melp,
1522c538 437 int devnum, char *metadata, int uuid[4], char *path);
ad5bc697
N
438extern int map_lock(struct map_ent **melp);
439extern void map_unlock(struct map_ent **melp);
cc700db3 440extern void map_fork(void);
e0d19036 441
e86c9dd6 442/* various details can be requested */
dab4a513
DW
443enum sysfs_read_flags {
444 GET_LEVEL = (1 << 0),
445 GET_LAYOUT = (1 << 1),
446 GET_COMPONENT = (1 << 2),
447 GET_CHUNK = (1 << 3),
448 GET_CACHE = (1 << 4),
449 GET_MISMATCH = (1 << 5),
450 GET_VERSION = (1 << 6),
451 GET_DISKS = (1 << 7),
452 GET_DEGRADED = (1 << 8),
453 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
454 GET_BITMAP_LOCATION = (1 << 10),
455
456 GET_DEVS = (1 << 20), /* gets role, major, minor */
457 GET_OFFSET = (1 << 21),
458 GET_SIZE = (1 << 22),
459 GET_STATE = (1 << 23),
460 GET_ERROR = (1 << 24),
dab4a513 461};
e86c9dd6
NB
462
463/* If fd >= 0, get the array it is open on,
464 * else use devnum. >=0 -> major9. <0.....
465 */
549e9569 466extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 467extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
468extern void sysfs_free(struct mdinfo *sra);
469extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
470extern int sysfs_attr_match(const char *attr, const char *str);
471extern int sysfs_match_word(const char *word, char **list);
7e0f6979 472extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 473 char *name, char *val);
7e0f6979 474extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 475 char *name, unsigned long long val);
97590376 476extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
477extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
478 char *name);
479extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 480extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 481 char *name, unsigned long long *val);
7236ee7a 482extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
483extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
484 char *name);
93ecfa01
N
485extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
486 char *name, char *val, int size);
8ed3e5e1 487extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 488extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 489extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 490extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 491extern int sysfs_unique_holder(int devnum, long rdev);
bc77ed53 492extern int sysfs_freeze_array(struct mdinfo *sra);
c69b251b 493extern int load_sys(char *path, char *buf);
130994cb
AK
494extern int reshape_prepare_fdlist(char *devname,
495 struct mdinfo *sra,
496 int raid_disks,
497 int nrdisks,
498 unsigned long blocks,
499 char *backup_file,
500 int *fdlist,
501 unsigned long long *offsets);
502extern void reshape_free_fdlist(int *fdlist,
503 unsigned long long *offsets,
504 int size);
e6e9d47b
AK
505extern int reshape_open_backup_file(char *backup,
506 int fd,
507 char *devname,
508 long blocks,
509 int *fdlist,
a93f87ee
N
510 unsigned long long *offsets,
511 int restart);
1c009fc2
AK
512extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
513 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
514
515extern int save_stripes(int *source, unsigned long long *offsets,
516 int raid_disks, int chunk_size, int level, int layout,
517 int nwrites, int *dest,
a6288483
N
518 unsigned long long start, unsigned long long length,
519 char *buf);
353632d9
NB
520extern int restore_stripes(int *dest, unsigned long long *offsets,
521 int raid_disks, int chunk_size, int level, int layout,
522 int source, unsigned long long read_offset,
2fcb75ae
AK
523 unsigned long long start, unsigned long long length,
524 char *src_buf);
e86c9dd6 525
52826846
NB
526#ifndef Sendmail
527#define Sendmail "/usr/lib/sendmail -t"
528#endif
529
773135f5
NB
530#define SYSLOG_FACILITY LOG_DAEMON
531
682c7051
NB
532extern char *map_num(mapping_t *map, int num);
533extern int map_name(mapping_t *map, char *name);
b640a252 534extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 535
c2ecf5f6
N
536extern char *map_dev_preferred(int major, int minor, int create,
537 char *prefer);
538static inline char *map_dev(int major, int minor, int create)
539{
540 return map_dev_preferred(major, minor, create, NULL);
541}
64c4757e 542
549e9569 543struct active_array;
2e735d19 544struct metadata_update;
64c4757e 545
999b4972
N
546
547/* 'struct reshape' records the intermediate states
548 * a general reshape.
549 * The starting geometry is converted to the 'before' geometry
550 * by at most an atomic level change. They could be the same.
551 * Similarly the 'after' geometry is converted to the final
552 * geometry by at most a level change.
553 * Note that 'before' and 'after' must have the same level.
554 * 'blocks' is the minimum number of sectors for a reshape unit.
555 * This will be a multiple of the stripe size in each of the
556 * 'before' and 'after' geometries.
557 * If 'blocks' is 0, no restriping is necessary.
558 */
559struct reshape {
560 int level;
561 int parity; /* number of parity blocks/devices */
562 struct {
563 int layout;
564 int data_disks;
565 } before, after;
566 unsigned long long backup_blocks;
567 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
568 unsigned long long new_size; /* New size of array in sectors */
569};
570
6adfd3af
NB
571/* A superswitch provides entry point the a metadata handler.
572 *
573 * The super_switch primarily operates on some "metadata" that
574 * is accessed via the 'supertype'.
575 * This metadata has one of three possible sources.
576 * 1/ It is read from a single device. In this case it may not completely
577 * describe the array or arrays as some information might be on other
578 * devices.
579 * 2/ It is read from all devices in a container. In this case all
580 * information is present.
581 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
582 * be complete once enough ->add_to_super calls have completed.
583 *
584 * When creating an array inside a container, the metadata will be
585 * formed by a combination of 2 and 3. The metadata or the array is read,
586 * then new information is added.
587 *
588 * The metadata must sometimes have a concept of a 'current' array
589 * and a 'current' device.
590 * The 'current' array is set by init_super to be the newly created array,
591 * or is set by super_by_fd when it finds it is looking at an array inside
592 * a container.
593 *
594 * The 'current' device is either the device that the metadata was read from
595 * in case 1, or the last device added by add_to_super in case 3.
596 * Case 2 does not identify a 'current' device.
597 */
f9ce90ba 598extern struct superswitch {
6adfd3af
NB
599
600 /* Used to report details of metadata read from a component
601 * device. ->load_super has been called.
602 */
3da92f27 603 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 604 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 605 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 606 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
607
608 /* Used to report details of an active array.
609 * ->load_super was possibly given a 'component' string.
610 */
3da92f27 611 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 612 void (*brief_detail_super)(struct supertype *st);
0d726f17 613 void (*export_detail_super)(struct supertype *st);
6adfd3af 614
4cce4069 615 /* Optional: platform hardware / firmware details */
5615172f 616 int (*detail_platform)(int verbose, int enumerate_only);
4cce4069 617
6adfd3af
NB
618 /* Used:
619 * to get uuid to storing in bitmap metadata
620 * and 'reshape' backup-data metadata
621 * To see if a device is being re-added to an array it was part of.
622 */
3da92f27 623 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 624
8592f29d 625 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
626 * the container, or about an individual array within the container.
627 * The determination is made either by:
628 * load_super being given a 'component' string.
629 * validate_geometry determining what to create.
d2ca6449
NB
630 * The info includes both array information and device information.
631 * The particular device should be:
632 * The last device added by add_to_super
633 * The device the metadata was loaded from by load_super
a5d85af7
N
634 * If 'map' is present, then it is an array raid_disks long
635 * (raid_disk must already be set and correct) and it is filled
636 * with 1 for slots that are thought to be active and 0 for slots which
637 * appear to be failed/missing.
95eeceeb 638 * *info is zeroed out before data is added.
6adfd3af 639 */
a5d85af7 640 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 641 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 642 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 643 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 644 */
3da92f27 645 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
646
647 /* Make one of several generic modifications to metadata
648 * prior to assembly (or other times).
649 * sparc2.2 - first bug in early 0.90 metadata
650 * super-minor - change name of 0.90 metadata
651 * summaries - 'correct' any redundant data
652 * resync - mark array as dirty to trigger a resync.
653 * uuid - set new uuid - only 0.90 or 1.x
654 * name - change the name of the array (where supported)
655 * homehost - change which host this array is tied to.
656 * devicesize - If metadata is at start of device, change recorded
657 * device size to match actual device size
658 * byteorder - swap bytes for 0.90 metadata
659 *
660 * force-one - mark that device as uptodate, not old or failed.
661 * force-array - mark array as clean if it would not otherwise
662 * assemble
663 * assemble - not sure how this is different from force-one...
664 * linear-grow-new - add a new device to a linear array, but don't
665 * change the size: so superblock still matches
666 * linear-grow-update - now change the size of the array.
16715c01
DL
667 * writemostly - set the WriteMostly1 bit in the superblock devflags
668 * readwrite - clear the WriteMostly1 bit in the superblock devflags
6adfd3af 669 */
68c7d6d7 670 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 671 char *update,
e5eac01f
NB
672 char *devname, int verbose,
673 int uuid_set, char *homehost);
6adfd3af
NB
674
675 /* Create new metadata for new array as described. This could
676 * be a new container, or an array in a pre-existing container.
677 * Also used to zero metadata prior to writing it to invalidate old
678 * metadata.
679 */
3da92f27
NB
680 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
681 unsigned long long size, char *name,
682 char *homehost, int *uuid);
6adfd3af
NB
683
684 /* update the metadata to include new device, either at create or
685 * when hot-adding a spare.
686 */
f20c3968 687 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
1a64be56
LM
688 int fd, char *devname);
689 /* update the metadata to delete a device,
690 * when hot-removing.
691 */
692 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
693
694 /* Write metadata to one device when fixing problems or adding
695 * a new device.
696 */
3da92f27 697 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
698
699 /* Write all metadata for this array.
700 */
111d01fc 701 int (*write_init_super)(struct supertype *st);
64557c33 702 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 703 int (*load_super)(struct supertype *st, int fd, char *devname);
2b959fbf 704 int (*load_container)(struct supertype *st, int fd, char *devname);
82d9eba6 705 struct supertype * (*match_metadata_desc)(char *arg);
1bf4e2d9 706 __u64 (*avail_size)(struct supertype *st, __u64 size);
80e7f8c3 707 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
3da92f27 708 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 709 int delay, int write_behind,
f9c25f1d 710 unsigned long long size, int may_change, int major);
3da92f27
NB
711 void (*locate_bitmap)(struct supertype *st, int fd);
712 int (*write_bitmap)(struct supertype *st, int fd);
713 void (*free_super)(struct supertype *st);
78e44928
NB
714
715 /* validate_geometry is called with an st returned by
716 * match_metadata_desc.
717 * It should check that the geometry described in compatible with
718 * the metadata type. It will be called repeatedly as devices
719 * added to validate changing size and new devices. If there are
720 * inter-device dependencies, it should record sufficient details
721 * so these can be validated.
30f58b22 722 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
723 * Return value is:
724 * 1: everything is OK
725 * 0: not OK for some reason - if 'verbose', then error was reported.
726 * -1: st->sb was NULL, 'subdev' is a member of a container of this
727 * types, but array is not acceptable for some reason
728 * message was reported even if verbose is 0.
78e44928 729 */
17f25ca6
NB
730 int (*validate_geometry)(struct supertype *st, int level, int layout,
731 int raiddisks,
c21e737b 732 int *chunk, unsigned long long size,
2c514b71
NB
733 char *subdev, unsigned long long *freesize,
734 int verbose);
598f0d58 735
00bbdbda 736 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
737 /* query the supertype for default geometry */
738 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
739 /* Permit subarray's to be deleted from inactive containers */
740 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 741 /* Permit subarray's to be modified */
a951a4f7 742 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 743 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
744 /* Check if reshape is supported for this external format.
745 * st is obtained from super_by_fd() where st->subarray[0] is
746 * initialized to indicate if reshape is being performed at the
747 * container or subarray level
748 */
016e00f5
AK
749#define APPLY_METADATA_CHANGES 1
750#define ROLLBACK_METADATA_CHANGES 0
751
7bc71196
DW
752 int (*reshape_super)(struct supertype *st, long long size, int level,
753 int layout, int chunksize, int raid_disks,
41784c88 754 int delta_disks, char *backup, char *dev,
016e00f5 755 int direction,
41784c88 756 int verbose); /* optional */
999b4972
N
757 int (*manage_reshape)( /* optional */
758 int afd, struct mdinfo *sra, struct reshape *reshape,
759 struct supertype *st, unsigned long blocks,
760 int *fds, unsigned long long *offsets,
761 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 762
549e9569 763/* for mdmon */
cba0191b
NB
764 int (*open_new)(struct supertype *c, struct active_array *a,
765 char *inst);
ed9d66aa
NB
766
767 /* Tell the metadata handler the current state of the array.
768 * This covers whether it is known to be consistent (no pending writes)
01f157d7 769 * and how far along a resync is known to have progressed
ed9d66aa
NB
770 * (in a->resync_start).
771 * resync status is really irrelevant if the array is not consistent,
772 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
773 * If 'consistent' is '2', then the array can mark it dirty if a
774 * resync/recovery/whatever is required, or leave it clean if not.
775 * Return value is 0 dirty (not consistent) and 1 if clean.
776 * it is only really important if consistent is passed in as '2'.
4e5528c6 777 */
01f157d7 778 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
779
780 /* When the state of a device might have changed, we call set_disk to
781 * tell the metadata what the current state is.
782 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
783 * transitions.
784 * set_disk might be called when the state of the particular disk has
785 * not in fact changed.
786 */
8d45d196 787 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
788 void (*sync_metadata)(struct supertype *st);
789 void (*process_update)(struct supertype *st,
790 struct metadata_update *update);
edd8d13c
NB
791 void (*prepare_update)(struct supertype *st,
792 struct metadata_update *update);
549e9569 793
6c3fb95c
NB
794 /* activate_spare will check if the array is degraded and, if it
795 * is, try to find some spare space in the container.
796 * On success, it add appropriate updates (For process_update) to
797 * to the 'updates' list and returns a list of 'mdinfo' identifying
798 * the device, or devices as there might be multiple missing
799 * devices and multiple spares available.
800 */
801 struct mdinfo *(*activate_spare)(struct active_array *a,
802 struct metadata_update **updates);
2cda7640
ML
803 /*
804 * Return statically allocated string that represents metadata specific
805 * controller domain of the disk. The domain is used in disk domain
806 * matching functions. Disks belong to the same domain if the they have
807 * the same domain from mdadm.conf and belong the same metadata domain.
808 * Returning NULL or not providing this handler means that metadata
809 * does not distinguish the differences between disks that belong to
810 * different controllers. They are in the domain specified by
811 * configuration file (mdadm.conf).
812 * In case when the metadata has the notion of domains based on disk
813 * it shall return NULL for disks that do not belong to the controller
814 * the supported domains. Such disks will form another domain and won't
815 * be mixed with supported ones.
816 */
817 const char *(*get_disk_controller_domain)(const char *path);
549e9569 818
ba53ea59
AK
819 /* for external backup area */
820 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
821
f277ce36 822 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 823 int external;
4cce4069 824 const char *name; /* canonical metadata name */
0f22b998 825} *superlist[];
82d9eba6 826
0f22b998
N
827extern struct superswitch super0, super1;
828extern struct superswitch super_imsm, super_ddf;
0592faeb 829extern struct superswitch mbr, gpt;
cdddbdbc 830
edd8d13c
NB
831struct metadata_update {
832 int len;
833 char *buf;
834 void *space; /* allocated space that monitor will use */
cb23f1f4
N
835 void **space_list; /* list of allocated spaces that monitor can
836 * use or that it returned.
837 */
edd8d13c
NB
838 struct metadata_update *next;
839};
840
6adfd3af
NB
841/* A supertype holds a particular collection of metadata.
842 * It identifies the metadata type by the superswitch, and the particular
843 * sub-version of that metadata type.
844 * metadata read in or created is stored in 'sb' and 'info'.
845 * There are also fields used by mdmon to track containers.
846 *
352452c3
N
847 * A supertype may refer to:
848 * Just an array, possibly in a container
849 * A container, not identifying any particular array
850 * Info read from just one device, not yet fully describing the array/container.
851 *
852 *
6adfd3af
NB
853 * A supertype is created by:
854 * super_by_fd
855 * guess_super
856 * dup_super
857 */
82d9eba6
NB
858struct supertype {
859 struct superswitch *ss;
860 int minor_version;
ea329559 861 int max_devs;
2f6079dc 862 int container_dev; /* devnum of container */
64557c33 863 void *sb;
111d01fc 864 void *info;
df3346e6
LM
865 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
866 HW/firmware related incompatability to load metadata.
867 Used when examining metadata to display content of disk
868 when user has no hw/firmare compatible system.
869 */
edd8d13c
NB
870 struct metadata_update *updates;
871 struct metadata_update **update_tail;
872
549e9569
NB
873 /* extra stuff used by mdmon */
874 struct active_array *arrays;
549e9569 875 int sock; /* listen to external programs */
549e9569
NB
876 int devnum;
877 char *devname; /* e.g. md0. This appears in metadata_verison:
878 * external:/md0/12
879 */
880 int devcnt;
68226a80 881 int retry_soon;
549e9569
NB
882
883 struct mdinfo *devs;
884
82d9eba6 885};
f9ce90ba 886
4725bc31 887extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
888enum guess_types { guess_any, guess_array, guess_partitions };
889extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
890static inline struct supertype *guess_super(int fd) {
891 return guess_super_type(fd, guess_any);
892}
3da92f27 893extern struct supertype *dup_super(struct supertype *st);
beae1dfe 894extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 895extern int must_be_container(int fd);
0f0749ad 896extern int dev_size_from_id(dev_t id, unsigned long long *size);
8382f19b
NB
897extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
898 mdu_disk_info_t *disk);
a7c6e3fb 899void wait_for(char *dev, int fd);
f9ce90ba 900
5527fc74
N
901/*
902 * Data structures for policy management.
903 * Each device can have a policy structure that lists
904 * various name/value pairs each possibly with a metadata associated.
905 * The policy list is sorted by name/value/metadata
906 */
907struct dev_policy {
908 struct dev_policy *next;
909 char *name; /* None of these strings are allocated. They are
910 * all just references to strings which are known
911 * to exist elsewhere.
912 * name and metadata can be compared by address equality.
913 */
914 const char *metadata;
2cda7640 915 const char *value;
5527fc74
N
916};
917
4e8d9f0a 918extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
919
920/* iterate over the sublist starting at list, having the same
921 * 'name' as 'list', and matching the given metadata (Where
922 * NULL matches anything
923 */
924#define pol_for_each(item, list, _metadata) \
925 for (item = list; \
926 item && item->name == list->name; \
927 item = item->next) \
928 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
929 ; else
930
931/*
932 * policy records read from mdadm are largely just name-value pairs.
933 * The names are constants, not strdupped
934 */
935struct pol_rule {
936 struct pol_rule *next;
937 char *type; /* rule_policy or rule_part */
938 struct rule {
939 struct rule *next;
940 char *name;
941 char *value;
942 char *dups; /* duplicates of 'value' with a partNN appended */
943 } *rule;
944};
945
946extern char rule_policy[], rule_part[];
947extern char rule_path[], rule_type[];
948extern char type_part[], type_disk[];
949
950extern void policyline(char *line, char *type);
4e8d9f0a 951extern void policy_add(char *type, ...);
5527fc74
N
952extern void policy_free(void);
953
73c9c47c 954extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 955extern struct dev_policy *disk_policy(struct mdinfo *disk);
4e8d9f0a 956extern struct dev_policy *devnum_policy(int dev);
5527fc74
N
957extern void dev_policy_free(struct dev_policy *p);
958
e78dda3b
N
959//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
960extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
961extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
962
e3bb5f14
N
963enum policy_action {
964 act_default,
965 act_include,
966 act_re_add,
d2db3045
N
967 act_spare, /* This only applies to bare devices */
968 act_spare_same_slot, /* this allows non-bare devices,
969 * but only if recent removal */
970 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
971 act_err
972};
973
974extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
975 enum policy_action want);
976extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
977 enum policy_action want);
f5f12c84
N
978
979struct domainlist {
980 struct domainlist *next;
2cda7640 981 const char *dom;
f5f12c84
N
982};
983
984extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
985 const char *metadata);
986extern struct domainlist *domain_from_array(struct mdinfo *mdi,
987 const char *metadata);
e78dda3b
N
988extern void domainlist_add_dev(struct domainlist **dom, int devnum,
989 const char *metadata);
f5f12c84
N
990extern void domain_free(struct domainlist *dl);
991extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
992 const char *metadata);
e78dda3b 993void domain_add(struct domainlist **domp, char *domain);
f5f12c84 994
403410eb
PC
995extern void policy_save_path(char *id_path, struct map_ent *array);
996extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
997
838acbc2
NB
998#if __GNUC__ < 3
999struct stat64;
1000#endif
1001
45e878bb
NB
1002#define HAVE_NFTW we assume
1003#define HAVE_FTW
1004
2df1f269 1005#ifdef __UCLIBC__
45e878bb 1006# include <features.h>
2df1f269
BRF
1007# ifndef __UCLIBC_HAS_LFS__
1008# define lseek64 lseek
1009# endif
45e878bb
NB
1010# ifndef __UCLIBC_HAS_FTW__
1011# undef HAVE_FTW
1012# undef HAVE_NFTW
1013# endif
1014#endif
1015
1016#ifdef __dietlibc__
1017# undef HAVE_NFTW
1018#endif
1019
e39b673e 1020#if defined(__KLIBC__)
1021# undef HAVE_NFTW
1022# undef HAVE_FTW
1023#endif
1024
45e878bb 1025#ifndef HAVE_NFTW
173fc515 1026# define FTW_PHYS 1
45e878bb
NB
1027# ifndef HAVE_FTW
1028 struct FTW {};
173fc515
NB
1029# endif
1030#endif
1031
45e878bb
NB
1032#ifdef HAVE_FTW
1033# include <ftw.h>
1034#endif
1035
173fc515
NB
1036extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1037
f9ce90ba 1038
64c4757e 1039extern int Manage_ro(char *devname, int fd, int readonly);
91f068bf 1040extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
dd0781e5 1041extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
64c4757e 1042extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1043 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1044 char *update, int force);
1f48664b 1045extern int autodetect(void);
e5329c37 1046extern int Grow_Add_device(char *devname, int fd, char *newdev);
8fac0577 1047extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
06b0d786 1048extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
e86c9dd6 1049 long long size,
691a36b7 1050 int level, char *layout_str, int chunksize, int raid_disks,
e2e53a2d 1051 struct mddev_dev *devlist,
ce52f92f 1052 int assume_clean, int force);
06b0d786 1053extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1054 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1055extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1056 struct mdinfo *info, char *backup_file,
1057 int freeze_reshape);
1058
3f54bd62
AK
1059extern int restore_backup(struct supertype *st,
1060 struct mdinfo *content,
1061 int working_disks,
1062 int spares,
1063 char *backup_file,
1064 int verbose);
2dddadb0
AK
1065extern int Grow_continue_command(char *devname, int fd,
1066 char *backup_file, int verbose);
64c4757e 1067
7f91af49 1068extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1069 struct mddev_ident *ident,
87f26d14
N
1070 struct mddev_dev *devlist,
1071 char *backup_file, int invalid_backup,
64c4757e 1072 int readonly, int runstop,
0ac91628 1073 char *update, char *homehost, int require_homehost,
b76b30e0 1074 int verbose, int force, int freeze_reshape);
64c4757e 1075
7f91af49 1076extern int Build(char *mddev, int chunk, int level, int layout,
a655e550 1077 int raiddisks, struct mddev_dev *devlist, int assume_clean,
7f91af49 1078 char *bitmap_file, int bitmap_chunk, int write_behind,
83208785 1079 int delay, int verbose, int autof, unsigned long long size);
64c4757e
NB
1080
1081
7f91af49 1082extern int Create(struct supertype *st, char *mddev,
5dd497ee 1083 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
3d3dd91e 1084 char *name, char *homehost, int *uuid,
a655e550 1085 int subdevs, struct mddev_dev *devlist,
47d79ef8 1086 int runstop, int verbose, int force, int assume_clean,
7f91af49 1087 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
64c4757e 1088
c2ecf5f6 1089extern int Detail(char *dev, int brief, int export, int test, char *homehost, char *prefer);
4cce4069 1090extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
e0d19036 1091extern int Query(char *dev);
a655e550 1092extern int Examine(struct mddev_dev *devlist, int brief, int export, int scan,
0d726f17 1093 int SparcAdjust, struct supertype *forcest, char *homehost);
a655e550 1094extern int Monitor(struct mddev_dev *devlist,
e0d19036 1095 char *mailaddr, char *alert_cmd,
aa88f531 1096 int period, int daemonise, int scan, int oneshot,
edde9560 1097 int dosyslog, int test, char *pidfile, int increments,
c2ecf5f6 1098 int share, char *prefer);
64c4757e 1099
9277cc77 1100extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
33414a01 1101extern int Kill_subarray(char *dev, char *subarray, int quiet);
fa56eddb 1102extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1103extern int Wait(char *dev);
9f1da824 1104extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1105
8382f19b 1106extern int Incremental(char *devname, int verbose, int runstop,
0ac91628 1107 struct supertype *st, char *homehost, int require_homehost,
b76b30e0 1108 int autof, int freeze_reshape);
8382f19b
NB
1109extern void RebuildMap(void);
1110extern int IncrementalScan(int verbose);
950bc344 1111extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1112extern int CreateBitmap(char *filename, int force, char uuid[16],
1113 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1114 unsigned long write_behind,
dcec9ee5
NB
1115 unsigned long long array_size,
1116 int major);
55935d51 1117extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1118extern int Write_rules(char *rule_name);
bf4fb153 1119extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1120extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 1121
64c4757e 1122extern int md_get_version(int fd);
e0d19036 1123extern int get_linux_version(void);
bc77ed53 1124extern int mdadm_version(char *version);
84e11361 1125extern long long parse_size(char *size);
64c4757e 1126extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1127extern int parse_layout_10(char *layout);
1128extern int parse_layout_faulty(char *layout);
682c7051
NB
1129extern int check_ext2(int fd, char *name);
1130extern int check_reiser(int fd, char *name);
1131extern int check_raid(int fd, char *name);
53ed6ac3
KW
1132extern int check_partitions(int fd, char *dname,
1133 unsigned long long freesize,
1134 unsigned long long size);
64c4757e 1135
dd0781e5 1136extern int get_mdp_major(void);
8b0dabea 1137extern int dev_open(char *dev, int flags);
e8a70c89 1138extern int open_dev(int devnum);
d998b738 1139extern int open_dev_flags(int devnum, int flags);
a931db9e 1140extern int open_dev_excl(int devnum);
8d80900b 1141extern int is_standard(char *dev, int *nump);
9008ed1c 1142extern int same_dev(char *one, char *two);
dd0781e5 1143
f1ae21c4 1144extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1145extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1146extern struct mddev_dev *conf_get_devs(void);
8382f19b 1147extern int conf_test_dev(char *devname);
4e8d9f0a 1148extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1149extern struct createinfo *conf_get_create_info(void);
1150extern void set_conffile(char *file);
1151extern char *conf_get_mailaddr(void);
1152extern char *conf_get_mailfrom(void);
1153extern char *conf_get_program(void);
0ac91628 1154extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1155extern char *conf_line(FILE *file);
98c6faba 1156extern char *conf_word(FILE *file, int allow_key);
0ac91628 1157extern int conf_name_is_free(char *name);
7c336758 1158extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1159extern int devname_matches(char *name, char *match);
2244d1a9
N
1160extern struct mddev_ident *conf_match(struct supertype *st,
1161 struct mdinfo *info,
1162 char *devname,
1163 int verbose, int *rvp);
6d11ec6f 1164extern int experimental(void);
0ac91628 1165
e0d19036
NB
1166extern void free_line(char *line);
1167extern int match_oneof(char *devices, char *devname);
e0d19036 1168extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1169extern const int uuid_zero[4];
f277ce36 1170extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1171extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1172extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1173extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1174 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1175extern unsigned long calc_csum(void *super, int bytes);
583315d9 1176extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1177 char *avail);
8453e704 1178extern int enough_fd(int fd);
e0d19036 1179extern int ask(char *mesg);
8fac0577 1180extern unsigned long long get_component_size(int fd);
0430ed48 1181extern void remove_partitions(int fd);
691c6ee1 1182extern int test_partition(int fd);
bfd76b93 1183extern int test_partition_from_id(dev_t id);
577e8448 1184extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1185extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1186 int chunksize, unsigned long long devsize);
edd8d13c
NB
1187extern int flush_metadata_updates(struct supertype *st);
1188extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c
DW
1189extern int assemble_container_content(struct supertype *st, int mdfd,
1190 struct mdinfo *content, int runstop,
49680258 1191 char *chosen_name, int verbose,
b76b30e0 1192 char *backup_file, int freeze_reshape);
326727d9
AC
1193extern struct mdinfo *container_choose_spares(struct supertype *st,
1194 unsigned long long min_size,
1195 struct domainlist *domlist,
1196 char *spare_group,
1197 const char *metadata, int get_one);
d52bb542 1198extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1199extern int add_disk(int mdfd, struct supertype *st,
1200 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1201extern int remove_disk(int mdfd, struct supertype *st,
1202 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1203extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1204unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1205
56eb10c0 1206extern char *human_size(long long bytes);
e4965ef8
N
1207extern char *human_size_brief(long long bytes);
1208extern void print_r10_layout(int layout);
e0d19036 1209
63152c1b
NB
1210#define NoMdDev (1<<23)
1211extern int find_free_devnum(int use_partitions);
1212
e0d19036
NB
1213extern void put_md_name(char *name);
1214extern char *get_md_name(int dev);
1215
c913b90e 1216extern char DefaultConfFile[];
b5e64645 1217
69207ff6
N
1218extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1219 char *chosen);
1220/* values for 'trustworthy' */
1221#define LOCAL 1
d1d3482b 1222#define LOCAL_ANY 10
69207ff6
N
1223#define FOREIGN 2
1224#define METADATA 3
6be1d39d 1225extern int open_mddev(char *dev, int report_errors);
a322f70c 1226extern int open_container(int fd);
33414a01
DW
1227extern int is_container_member(struct mdstat_ent *ent, char *devname);
1228extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1229extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1230extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1231
a931db9e 1232extern int mdmon_running(int devnum);
24f6f99b 1233extern int mdmon_pid(int devnum);
40ebbb9c 1234extern int check_env(char *name);
148acb7b 1235extern __u32 random32(void);
8850ee3e 1236extern int start_mdmon(int devnum);
a931db9e 1237
999b4972
N
1238extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1239 struct supertype *st, unsigned long stripes,
1240 int *fds, unsigned long long *offsets,
1241 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1242void abort_reshape(struct mdinfo *sra);
999b4972 1243
2f6079dc 1244extern char *devnum2devname(int num);
11877f4d 1245extern void fmt_devname(char *name, int num);
77472ff8 1246extern int devname2devnum(char *name);
c94709e8 1247extern int stat2devnum(struct stat *st);
2f6079dc 1248extern int fd2devnum(int fd);
b5e64645 1249
a931db9e
NB
1250static inline int dev2major(int d)
1251{
1252 if (d >= 0)
1253 return MD_MAJOR;
1254 else
1255 return get_mdp_major();
1256}
1257
1258static inline int dev2minor(int d)
1259{
1260 if (d >= 0)
1261 return d;
1262 return (-1-d) << MdpMinorShift;
1263}
1264
2d762ade 1265#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1266#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1267#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1268 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1269
3c558363
N
1270static inline int is_subarray(char *vers)
1271{
e9dd1598
N
1272 /* The version string for a 'subarray' (an array in a container)
1273 * is
1274 * /containername/componentname for normal read-write arrays
746a6567
N
1275 * -containername/componentname for arrays which mdmon must not
1276 * reconfigure. They might be read-only
1277 * or might be undergoing reshape etc.
e9dd1598
N
1278 * containername is e.g. md0, md_d1
1279 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1280 */
1281 return (*vers == '/' || *vers == '-');
3c558363
N
1282}
1283
e5408a32
DW
1284static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1285{
1286 return &ent->metadata_version[10+strlen(container)+1];
1287}
1288
2a24d7b6
DW
1289#ifdef DEBUG
1290#define dprintf(fmt, arg...) \
1291 fprintf(stderr, fmt, ##arg)
1292#else
1293#define dprintf(fmt, arg...) \
1294 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1295#endif
1a0ee0ba
DK
1296#include <assert.h>
1297#include <stdarg.h>
1298static inline int xasprintf(char **strp, const char *fmt, ...) {
1299 va_list ap;
1300 int ret;
1301 va_start(ap, fmt);
089485cb 1302 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1303 va_end(ap);
1304 assert(ret >= 0);
1305 return ret;
1306}
2a24d7b6 1307
b5e64645
NB
1308#define LEVEL_MULTIPATH (-4)
1309#define LEVEL_LINEAR (-1)
1310#define LEVEL_FAULTY (-5)
1311
17f25ca6
NB
1312/* kernel module doesn't know about these */
1313#define LEVEL_CONTAINER (-100)
a322f70c 1314#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1315
b5e64645
NB
1316
1317/* faulty stuff */
1318
1319#define WriteTransient 0
1320#define ReadTransient 1
1321#define WritePersistent 2
1322#define ReadPersistent 3
1323#define WriteAll 4 /* doesn't go to device */
1324#define ReadFixable 5
1325#define Modes 6
1326
1327#define ClearErrors 31
1328#define ClearFaults 30
1329
1330#define AllPersist 100 /* internal use only */
1331#define NoPersist 101
1332
1333#define ModeMask 0x1f
1334#define ModeShift 5
fe6729fa
NB
1335
1336
1337#ifdef __TINYC__
1338#undef minor
1339#undef major
1340#undef makedev
1341#define minor(x) ((x)&0xff)
1342#define major(x) (((x)>>8)&0xff)
1343#define makedev(M,m) (((M)<<8) | (m))
1344#endif
1345
b640a252 1346/* for raid4/5/6 */
e86c9dd6
NB
1347#define ALGORITHM_LEFT_ASYMMETRIC 0
1348#define ALGORITHM_RIGHT_ASYMMETRIC 1
1349#define ALGORITHM_LEFT_SYMMETRIC 2
1350#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1351
1352/* Define non-rotating (raid4) algorithms. These allow
1353 * conversion of raid4 to raid5.
1354 */
1355#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1356#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1357
1358/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1359 * Firstly, the exact positioning of the parity block is slightly
1360 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1361 * of DDF.
1362 * Secondly, or order of datablocks over which the Q syndrome is computed
1363 * is different.
1364 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1365 * These layouts are from the DDFv1.2 spec.
1366 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1367 * leaves RLQ=3 as 'Vendor Specific'
1368 */
1369
1370#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1371#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1372#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1373
1374
1375/* For every RAID5 algorithm we define a RAID6 algorithm
1376 * with exactly the same layout for data and parity, and
1377 * with the Q block always on the last device (N-1).
1378 * This allows trivial conversion from RAID5 to RAID6
1379 */
1380#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1381#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1382#define ALGORITHM_LEFT_SYMMETRIC_6 18
1383#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1384#define ALGORITHM_PARITY_0_6 20
1385#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1386
33a6535d
AW
1387/* Define PATH_MAX in case we don't use glibc or standard library does
1388 * not have PATH_MAX defined. Assume max path length is 4K characters.
1389 */
1390#ifndef PATH_MAX
1391#define PATH_MAX 4096
1392#endif
1393
2d3603ba
KW
1394#define PROCESS_DELAYED -2
1395#define PROCESS_PENDING -3
a0963a86 1396
480f3566
N
1397/* When using "GET_DISK_INFO" it isn't certain how high
1398 * we need to check. So we impose an absolute limit of
1399 * MAX_DISKS. This needs to be much more than the largest
1400 * number of devices any metadata can support. Currently
1401 * v1.x can support 1920
1402 */
1403#define MAX_DISKS 4096
1404
a0963a86 1405extern int __offroot;