]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Fix typos in mdadm.8.in
[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>
0d35d5c4 28#ifdef __GLIBC__
64c4757e 29extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
0d35d5c4 30#elif !defined(lseek64)
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 54#include <linux/kdev_t.h>
e0d19036
NB
55/*#include <linux/fs.h> */
56#include <sys/mount.h>
57#include <asm/types.h>
64c4757e
NB
58#include <sys/ioctl.h>
59#define MD_MAJOR 9
dd0781e5 60#define MdpMinorShift 6
64c4757e 61
e0d19036 62#ifndef BLKGETSIZE64
98c6faba 63#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 64#endif
56eb10c0 65
c21e737b 66#define DEFAULT_CHUNK 512
c82f047c
NB
67#define DEFAULT_BITMAP_CHUNK 4096
68#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 69#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e 70
753cf905 71/* MAP_DIR should be somewhere that persists across the pivotroot
5d4d1b26 72 * from early boot to late boot.
96fd06ed 73 * /run seems to have emerged as the best standard.
5d4d1b26 74 */
753cf905 75#ifndef MAP_DIR
96fd06ed 76#define MAP_DIR "/run/mdadm"
753cf905
DL
77#endif /* MAP_DIR */
78/* MAP_FILE is what we name the map file we put in MAP_DIR, in case you
79 * want something other than the default of "map"
80 */
81#ifndef MAP_FILE
82#define MAP_FILE "map"
83#endif /* MAP_FILE */
84/* MDMON_DIR is where pid and socket files used for communicating
96fd06ed 85 * with mdmon normally live. Best is /var/run/mdadm as
753cf905
DL
86 * mdmon is needed at early boot then it needs to write there prior
87 * to /var/run being mounted read/write, and it also then needs to
88 * persist beyond when /var/run is mounter read-only. So, to be
89 * safe, the default is somewhere that is read/write early in the
90 * boot process and stays up as long as possible during shutdown.
91 */
92#ifndef MDMON_DIR
96fd06ed 93#define MDMON_DIR "/run/mdadm"
753cf905 94#endif /* MDMON_DIR */
5d4d1b26 95
403410eb
PC
96/* FAILED_SLOTS is where to save files storing recent removal of array
97 * member in order to allow future reuse of disk inserted in the same
98 * slot for array recovery
99 */
100#ifndef FAILED_SLOTS_DIR
96fd06ed 101#define FAILED_SLOTS_DIR "/run/mdadm/failed-slots"
403410eb
PC
102#endif /* FAILED_SLOTS */
103
64c4757e 104#include "md_u.h"
e0d19036 105#include "md_p.h"
c82f047c 106#include "bitmap.h"
f7dd881f 107#include "msg.h"
64c4757e 108
1e0d770c 109#include <endian.h>
efd441d1
NB
110/* Redhat don't like to #include <asm/byteorder.h>, and
111 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
112 * and there is no standard conversion function so... */
0ae03b8a
NB
113/* And dietlibc doesn't think byteswap is ok, so.. */
114/* #include <byteswap.h> */
115#define bswap_16(x) (((x) & 0x00ffU) << 8 | \
116 ((x) & 0xff00U) >> 8)
117#define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
118 ((x) & 0xff000000U) >> 24 | \
119 ((x) & 0x0000ff00U) << 8 | \
120 ((x) & 0x00ff0000U) >> 8)
121#define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
122 ((x) & 0xff00000000000000ULL) >> 56 | \
123 ((x) & 0x000000000000ff00ULL) << 40 | \
124 ((x) & 0x00ff000000000000ULL) >> 40 | \
125 ((x) & 0x0000000000ff0000ULL) << 24 | \
126 ((x) & 0x0000ff0000000000ULL) >> 24 | \
127 ((x) & 0x00000000ff000000ULL) << 8 | \
128 ((x) & 0x000000ff00000000ULL) >> 8)
129
e39b673e 130#if !defined(__KLIBC__)
efd441d1 131#if BYTE_ORDER == LITTLE_ENDIAN
8a5f247d
JS
132#define __cpu_to_le16(_x) (unsigned int)(_x)
133#define __cpu_to_le32(_x) (unsigned int)(_x)
134#define __cpu_to_le64(_x) (unsigned long long)(_x)
135#define __le16_to_cpu(_x) (unsigned int)(_x)
136#define __le32_to_cpu(_x) (unsigned int)(_x)
137#define __le64_to_cpu(_x) (unsigned long long)(_x)
974e620d
NB
138
139#define __cpu_to_be16(_x) bswap_16(_x)
140#define __cpu_to_be32(_x) bswap_32(_x)
141#define __cpu_to_be64(_x) bswap_64(_x)
142#define __be16_to_cpu(_x) bswap_16(_x)
143#define __be32_to_cpu(_x) bswap_32(_x)
144#define __be64_to_cpu(_x) bswap_64(_x)
efd441d1
NB
145#elif BYTE_ORDER == BIG_ENDIAN
146#define __cpu_to_le16(_x) bswap_16(_x)
147#define __cpu_to_le32(_x) bswap_32(_x)
148#define __cpu_to_le64(_x) bswap_64(_x)
149#define __le16_to_cpu(_x) bswap_16(_x)
150#define __le32_to_cpu(_x) bswap_32(_x)
151#define __le64_to_cpu(_x) bswap_64(_x)
974e620d 152
8a5f247d
JS
153#define __cpu_to_be16(_x) (unsigned int)(_x)
154#define __cpu_to_be32(_x) (unsigned int)(_x)
155#define __cpu_to_be64(_x) (unsigned long long)(_x)
156#define __be16_to_cpu(_x) (unsigned int)(_x)
157#define __be32_to_cpu(_x) (unsigned int)(_x)
158#define __be64_to_cpu(_x) (unsigned long long)(_x)
1e0d770c
NB
159#else
160# error "unknown endianness."
161#endif
e39b673e 162#endif /* __KLIBC__ */
1e0d770c 163
1e5c6983
DW
164/*
165 * min()/max()/clamp() macros that also do
166 * strict type-checking.. See the
167 * "unnecessary" pointer comparison.
168 */
169#define min(x, y) ({ \
170 typeof(x) _min1 = (x); \
171 typeof(y) _min2 = (y); \
172 (void) (&_min1 == &_min2); \
173 _min1 < _min2 ? _min1 : _min2; })
174
175#define max(x, y) ({ \
176 typeof(x) _max1 = (x); \
177 typeof(y) _max2 = (y); \
178 (void) (&_max1 == &_max2); \
179 _max1 > _max2 ? _max1 : _max2; })
1e0d770c 180
1c0aebc2
N
181#define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))
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;
fe384ca0 191 unsigned long long new_data_offset;
598f0d58
NB
192 unsigned long long component_size; /* same as array.size, except in
193 * sectors and up to 64bits.
194 */
da9b4a62
DW
195 unsigned long long custom_array_size; /* size for non-default sized
196 * arrays (in sectors)
197 */
81219e70
LM
198#define NO_RESHAPE 0
199#define VOLUME_RESHAPE 1
200#define CONTAINER_RESHAPE 2
5e88ab2e 201#define RESHAPE_NO_BACKUP 16 /* Mask 'or'ed in */
353632d9
NB
202 int reshape_active;
203 unsigned long long reshape_progress;
6e75048b
AK
204 int recovery_blocked; /* for external metadata it
205 * indicates that there is
206 * reshape in progress in
207 * container,
208 * for native metadata it is
209 * reshape_active field mirror
210 */
80bf9135
N
211 /* During reshape we can sometimes change the data_offset to avoid
212 * over-writing still-valid data. We need to know if there is space.
213 * So getinfo_super will fill in space_before and space_after in sectors.
214 * data_offset can be increased or decreased by this amount.
215 */
216 unsigned long long space_before, space_after;
e1516be1
DW
217 union {
218 unsigned long long resync_start; /* per-array resync position */
219 unsigned long long recovery_start; /* per-device rebuild position */
b7528a20 220 #define MaxSector (~0ULL) /* resync/recovery complete position */
e1516be1 221 };
fbdef498 222 long bitmap_offset; /* 0 == none, 1 == a file */
a67dd8cc 223 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 224 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 225 int errors;
f21e18ca 226 unsigned long cache_size; /* size of raid456 stripe cache*/
7e0f6979
NB
227 int mismatch_cnt;
228 char text_version[50];
dd15dc4a
NB
229
230 int container_member; /* for assembling external-metatdata arrays
231 * This is to be used internally by metadata
232 * handler only */
97b4d0e9
DW
233 int container_enough; /* flag external handlers can set to
234 * indicate that subarrays have not enough (-1),
235 * enough to start (0), or all expected disks (1) */
1011e834 236 char sys_name[20];
7e0f6979 237 struct mdinfo *devs;
06c7f68e 238 struct mdinfo *next;
549e9569
NB
239
240 /* Device info for mdmon: */
e1516be1 241 int recovery_fd;
549e9569 242 int state_fd;
8d45d196
DW
243 #define DS_FAULTY 1
244 #define DS_INSYNC 2
245 #define DS_WRITE_MOSTLY 4
246 #define DS_SPARE 8
247 #define DS_BLOCKED 16
248 #define DS_REMOVE 1024
92967543 249 #define DS_UNBLOCK 2048
549e9569
NB
250 int prev_state, curr_state, next_state;
251
4b1ac34b
NB
252};
253
5bbb4842
NB
254struct createinfo {
255 int uid;
256 int gid;
257 int autof;
258 int mode;
38098016 259 int symlinks;
eca944fa 260 int names;
058574b1 261 struct supertype *supertype;
5bbb4842
NB
262};
263
9a9dab36 264#define Name "mdadm"
682c7051 265
e0d19036
NB
266enum mode {
267 ASSEMBLE=1,
268 BUILD,
269 CREATE,
270 MANAGE,
271 MISC,
272 MONITOR,
dd0781e5 273 GROW,
8382f19b 274 INCREMENTAL,
1f48664b 275 AUTODETECT,
5187a385 276 mode_count
e0d19036
NB
277};
278
64c4757e 279extern char short_options[];
024768c4 280extern char short_bitmap_options[];
c06487ce 281extern char short_bitmap_auto_options[];
64c4757e 282extern struct option long_options[];
56eedc1a 283extern char Version[], Usage[], Help[], OptionHelp[],
5187a385 284 *mode_help[],
dd0781e5 285 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 286 Help_incr[],
e0d19036 287 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 288
997aed5d 289/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 290 * numbers later than any 'short' character option.
997aed5d
NB
291 */
292enum special_options {
1c7a808c 293 AssumeClean = 300,
997aed5d
NB
294 BitmapChunk,
295 WriteBehind,
296 ReAdd,
297 NoDegraded,
298 Sparc22,
1c7a808c 299 BackupFile,
997aed5d 300 HomeHost,
589395d6 301 AutoHomeHost,
38098016 302 Symlinks,
1f48664b 303 AutoDetect,
1770662b 304 Waitclean,
4cce4069 305 DetailPlatform,
33414a01 306 KillSubarray,
1c7a808c 307 UpdateSubarray,
edde9560 308 IncrementalPath,
1c7a808c
N
309 NoSharing,
310 HelpOptions,
311 Brief,
312 ManageOpt,
313 Add,
314 Remove,
315 Fail,
70c55e36
N
316 Replace,
317 With,
1c7a808c
N
318 MiscOpt,
319 WaitOpt,
320 ConfigFile,
321 ChunkSize,
322 WriteMostly,
323 Layout,
324 Auto,
325 Force,
326 SuperMinor,
327 EMail,
328 ProgramOpt,
329 Increment,
330 Fork,
331 Bitmap,
332 RebuildMapOpt,
87f26d14 333 InvalidBackup,
20b60dcd 334 UdevRules,
b76b30e0 335 FreezeReshape,
2dddadb0 336 Continue,
08ca2adf 337 OffRootOpt,
c2ecf5f6 338 Prefer,
f7d3febc 339 KillOpt,
40c9a66a 340 DataOffset,
6d388a88 341 ExamineBB,
74db60b0
N
342 Dump,
343 Restore,
997aed5d
NB
344};
345
f0ec6710
MN
346enum prefix_standard {
347 JEDEC,
348 IEC
349};
350
64c4757e 351/* structures read from config file */
52826846
NB
352/* List of mddevice names and identifiers
353 * Identifiers can be:
354 * uuid=128-hex-uuid
355 * super-minor=decimal-minor-number-from-superblock
356 * devices=comma,separated,list,of,device,names,with,wildcards
357 *
358 * If multiple fields are present, the intersection of all matching
359 * devices is considered
360 */
98c6faba 361#define UnSet (0xfffe)
fa56eddb 362struct mddev_ident {
dd0781e5 363 char *devname;
aba69144 364
dd0781e5 365 int uuid_set;
3fa06e9d 366 int uuid[4];
947fd4dd 367 char name[33];
52826846 368
f21e18ca 369 int super_minor;
52826846 370
dd0781e5 371 char *devices; /* comma separated list of device
52826846
NB
372 * names with wild cards
373 */
dd0781e5 374 int level;
f21e18ca
N
375 int raid_disks;
376 int spare_disks;
82d9eba6 377 struct supertype *st;
dd0781e5
NB
378 int autof; /* 1 for normal, 2 for partitioned */
379 char *spare_group;
7ef02d01 380 char *bitmap_file;
c82f047c 381 int bitmap_fd;
dd0781e5 382
1771a6e2
N
383 char *container; /* /dev/whatever name of container, or
384 * uuid of container. You would expect
385 * this to be the 'devname' or UUID
dbb44303
N
386 * of some other entry.
387 */
388 char *member; /* subarray within a container */
389
fa56eddb 390 struct mddev_ident *next;
b179246f
N
391 union {
392 /* fields needed by different users of this structure */
393 int assembled; /* set when assembly succeeds */
394 };
fa56eddb 395};
64c4757e 396
9e33d556
N
397struct context {
398 int readonly;
399 int runstop;
400 int verbose;
9e33d556
N
401 int brief;
402 int force;
403 char *homehost;
404 int require_homehost;
405 char *prefer;
406 int export;
407 int test;
408 char *subarray;
409 char *update;
410 int scan;
411 int SparcAdjust;
412 int autof;
413 int delay;
414 int freeze_reshape;
415 char *backup_file;
416 int invalid_backup;
417};
418
e705e81b
N
419struct shape {
420 int raiddisks;
421 int sparedisks;
422 int level;
423 int layout;
424 char *layout_str;
425 int chunk;
426 int bitmap_chunk;
427 char *bitmap_file;
428 int assume_clean;
429 int write_behind;
430 unsigned long long size;
431};
432
64c4757e 433/* List of device names - wildcards expanded */
a655e550 434struct mddev_dev {
64c4757e 435 char *devname;
c8e1a230
N
436 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail,
437 * 'A' for re_add.
cd29a5c8
NB
438 * Not set for names read from .config
439 */
b3d31955 440 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
70c55e36 441 int used; /* set when used */
72ca9bcf 442 long long data_offset;
a655e550
N
443 struct mddev_dev *next;
444};
64c4757e 445
682c7051
NB
446typedef struct mapping {
447 char *name;
448 int num;
449} mapping_t;
450
e0d19036
NB
451struct mdstat_ent {
452 char *dev;
4dd2df09 453 char devnm[32];
e0d19036
NB
454 int active;
455 char *level;
456 char *pattern; /* U or up, _ for down */
457 int percent; /* -1 if no resync */
f94c116f 458 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
459 int devcnt;
460 int raid_disks;
549e9569 461 char * metadata_version;
3b57c466
N
462 struct dev_member {
463 char *name;
464 struct dev_member *next;
1011e834 465 } *members;
e0d19036
NB
466 struct mdstat_ent *next;
467};
468
22a88995 469extern struct mdstat_ent *mdstat_read(int hold, int start);
a7a0d8a1 470extern void mdstat_close(void);
e0d19036 471extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 472extern void mdstat_wait(int seconds);
58a4ba2a 473extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
4dd2df09 474extern int mddev_busy(char *devnm);
3b57c466 475extern struct mdstat_ent *mdstat_by_component(char *name);
4dd2df09 476extern struct mdstat_ent *mdstat_by_subdev(char *subdev, char *container);
8382f19b
NB
477
478struct map_ent {
479 struct map_ent *next;
4dd2df09 480 char devnm[32];
1522c538 481 char metadata[20];
8382f19b 482 int uuid[4];
195254b8 483 int bad;
8382f19b
NB
484 char *path;
485};
4dd2df09 486extern int map_update(struct map_ent **mpp, char *devnm, char *metadata,
8382f19b 487 int uuid[4], char *path);
4dd2df09 488extern void map_remove(struct map_ent **map, char *devnm);
8382f19b 489extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
eb2306f8
N
490#ifdef MDASSEMBLE
491static inline struct map_ent *map_by_devnm(struct map_ent **map, char *name)
492{
493 return NULL;
494}
495static inline void map_free(struct map_ent *map)
496{
497}
498#else
4dd2df09 499extern struct map_ent *map_by_devnm(struct map_ent **map, char *devnm);
eb2306f8
N
500extern void map_free(struct map_ent *map);
501#endif
f2e55ecc 502extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
503extern void map_read(struct map_ent **melp);
504extern int map_write(struct map_ent *mel);
4dd2df09 505extern void map_delete(struct map_ent **mapp, char *devnm);
8382f19b 506extern void map_add(struct map_ent **melp,
4dd2df09 507 char *devnm, char *metadata, int uuid[4], char *path);
ad5bc697
N
508extern int map_lock(struct map_ent **melp);
509extern void map_unlock(struct map_ent **melp);
cc700db3 510extern void map_fork(void);
e0d19036 511
e86c9dd6 512/* various details can be requested */
dab4a513
DW
513enum sysfs_read_flags {
514 GET_LEVEL = (1 << 0),
515 GET_LAYOUT = (1 << 1),
516 GET_COMPONENT = (1 << 2),
517 GET_CHUNK = (1 << 3),
518 GET_CACHE = (1 << 4),
519 GET_MISMATCH = (1 << 5),
520 GET_VERSION = (1 << 6),
521 GET_DISKS = (1 << 7),
522 GET_DEGRADED = (1 << 8),
523 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
524 GET_BITMAP_LOCATION = (1 << 10),
525
526 GET_DEVS = (1 << 20), /* gets role, major, minor */
527 GET_OFFSET = (1 << 21),
528 GET_SIZE = (1 << 22),
529 GET_STATE = (1 << 23),
530 GET_ERROR = (1 << 24),
dab4a513 531};
e86c9dd6
NB
532
533/* If fd >= 0, get the array it is open on,
4dd2df09 534 * else use devnm.
e86c9dd6 535 */
4dd2df09
N
536extern int sysfs_open(char *devnm, char *devname, char *attr);
537extern void sysfs_init(struct mdinfo *mdi, int fd, char *devnm);
7e0f6979 538extern void sysfs_free(struct mdinfo *sra);
4dd2df09 539extern struct mdinfo *sysfs_read(int fd, char *devnm, unsigned long options);
1770662b
DW
540extern int sysfs_attr_match(const char *attr, const char *str);
541extern int sysfs_match_word(const char *word, char **list);
7e0f6979 542extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 543 char *name, char *val);
7e0f6979 544extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 545 char *name, unsigned long long val);
012a8641
JS
546extern int sysfs_set_num_signed(struct mdinfo *sra, struct mdinfo *dev,
547 char *name, long long val);
97590376 548extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
549extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
550 char *name);
551extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 552extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 553 char *name, unsigned long long *val);
2eba8496
N
554extern int sysfs_fd_get_two(int fd, unsigned long long *v1, unsigned long long *v2);
555extern int sysfs_get_two(struct mdinfo *sra, struct mdinfo *dev,
556 char *name, unsigned long long *v1, unsigned long long *v2);
7236ee7a 557extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
558extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
559 char *name);
93ecfa01
N
560extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
561 char *name, char *val, int size);
8ed3e5e1 562extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 563extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 564extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 565extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
4dd2df09 566extern int sysfs_unique_holder(char *devnm, long rdev);
bc77ed53 567extern int sysfs_freeze_array(struct mdinfo *sra);
efc67e8e 568extern int sysfs_wait(int fd, int *msec);
c69b251b 569extern int load_sys(char *path, char *buf);
130994cb
AK
570extern int reshape_prepare_fdlist(char *devname,
571 struct mdinfo *sra,
572 int raid_disks,
573 int nrdisks,
574 unsigned long blocks,
575 char *backup_file,
576 int *fdlist,
577 unsigned long long *offsets);
578extern void reshape_free_fdlist(int *fdlist,
579 unsigned long long *offsets,
580 int size);
e6e9d47b
AK
581extern int reshape_open_backup_file(char *backup,
582 int fd,
583 char *devname,
584 long blocks,
585 int *fdlist,
a93f87ee
N
586 unsigned long long *offsets,
587 int restart);
1c009fc2
AK
588extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
589 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
590
591extern int save_stripes(int *source, unsigned long long *offsets,
592 int raid_disks, int chunk_size, int level, int layout,
593 int nwrites, int *dest,
a6288483
N
594 unsigned long long start, unsigned long long length,
595 char *buf);
353632d9
NB
596extern int restore_stripes(int *dest, unsigned long long *offsets,
597 int raid_disks, int chunk_size, int level, int layout,
598 int source, unsigned long long read_offset,
2fcb75ae
AK
599 unsigned long long start, unsigned long long length,
600 char *src_buf);
e86c9dd6 601
52826846
NB
602#ifndef Sendmail
603#define Sendmail "/usr/lib/sendmail -t"
604#endif
605
773135f5
NB
606#define SYSLOG_FACILITY LOG_DAEMON
607
682c7051
NB
608extern char *map_num(mapping_t *map, int num);
609extern int map_name(mapping_t *map, char *name);
b640a252 610extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 611
c2ecf5f6
N
612extern char *map_dev_preferred(int major, int minor, int create,
613 char *prefer);
614static inline char *map_dev(int major, int minor, int create)
615{
616 return map_dev_preferred(major, minor, create, NULL);
617}
64c4757e 618
549e9569 619struct active_array;
2e735d19 620struct metadata_update;
64c4757e 621
80bf9135 622/* 'struct reshape' records the intermediate states of
999b4972
N
623 * a general reshape.
624 * The starting geometry is converted to the 'before' geometry
625 * by at most an atomic level change. They could be the same.
626 * Similarly the 'after' geometry is converted to the final
627 * geometry by at most a level change.
628 * Note that 'before' and 'after' must have the same level.
629 * 'blocks' is the minimum number of sectors for a reshape unit.
630 * This will be a multiple of the stripe size in each of the
631 * 'before' and 'after' geometries.
632 * If 'blocks' is 0, no restriping is necessary.
89ecd3cf
N
633 * 'min_offset_change' is the minimum change to data_offset to
634 * allow the reshape to happen. It is at least the larger of
635 * the old and new chunk sizes, and typically the same as 'blocks'
636 * divided by number of data disks.
999b4972
N
637 */
638struct reshape {
639 int level;
640 int parity; /* number of parity blocks/devices */
641 struct {
642 int layout;
643 int data_disks;
644 } before, after;
645 unsigned long long backup_blocks;
89ecd3cf 646 unsigned long long min_offset_change;
999b4972
N
647 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
648 unsigned long long new_size; /* New size of array in sectors */
649};
650
6adfd3af
NB
651/* A superswitch provides entry point the a metadata handler.
652 *
74db60b0 653 * The superswitch primarily operates on some "metadata" that
6adfd3af
NB
654 * is accessed via the 'supertype'.
655 * This metadata has one of three possible sources.
656 * 1/ It is read from a single device. In this case it may not completely
657 * describe the array or arrays as some information might be on other
658 * devices.
659 * 2/ It is read from all devices in a container. In this case all
660 * information is present.
661 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
662 * be complete once enough ->add_to_super calls have completed.
663 *
664 * When creating an array inside a container, the metadata will be
665 * formed by a combination of 2 and 3. The metadata or the array is read,
666 * then new information is added.
667 *
668 * The metadata must sometimes have a concept of a 'current' array
669 * and a 'current' device.
670 * The 'current' array is set by init_super to be the newly created array,
671 * or is set by super_by_fd when it finds it is looking at an array inside
672 * a container.
673 *
674 * The 'current' device is either the device that the metadata was read from
675 * in case 1, or the last device added by add_to_super in case 3.
676 * Case 2 does not identify a 'current' device.
677 */
f9ce90ba 678extern struct superswitch {
6adfd3af
NB
679
680 /* Used to report details of metadata read from a component
681 * device. ->load_super has been called.
682 */
3da92f27 683 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 684 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 685 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 686 void (*export_examine_super)(struct supertype *st);
6d388a88 687 int (*examine_badblocks)(struct supertype *st, int fd, char *devname);
74db60b0 688 int (*copy_metadata)(struct supertype *st, int from, int to);
6adfd3af
NB
689
690 /* Used to report details of an active array.
691 * ->load_super was possibly given a 'component' string.
692 */
3da92f27 693 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 694 void (*brief_detail_super)(struct supertype *st);
0d726f17 695 void (*export_detail_super)(struct supertype *st);
6adfd3af 696
4cce4069 697 /* Optional: platform hardware / firmware details */
9eafa1de
MN
698 int (*detail_platform)(int verbose, int enumerate_only, char *controller_path);
699 int (*export_detail_platform)(int verbose, char *controller_path);
4cce4069 700
6adfd3af
NB
701 /* Used:
702 * to get uuid to storing in bitmap metadata
703 * and 'reshape' backup-data metadata
704 * To see if a device is being re-added to an array it was part of.
705 */
3da92f27 706 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 707
8592f29d 708 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
709 * the container, or about an individual array within the container.
710 * The determination is made either by:
711 * load_super being given a 'component' string.
712 * validate_geometry determining what to create.
d2ca6449
NB
713 * The info includes both array information and device information.
714 * The particular device should be:
715 * The last device added by add_to_super
716 * The device the metadata was loaded from by load_super
a5d85af7
N
717 * If 'map' is present, then it is an array raid_disks long
718 * (raid_disk must already be set and correct) and it is filled
719 * with 1 for slots that are thought to be active and 0 for slots which
720 * appear to be failed/missing.
95eeceeb 721 * *info is zeroed out before data is added.
6adfd3af 722 */
a5d85af7 723 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 724 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 725 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 726 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 727 */
3da92f27 728 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
729
730 /* Make one of several generic modifications to metadata
731 * prior to assembly (or other times).
732 * sparc2.2 - first bug in early 0.90 metadata
733 * super-minor - change name of 0.90 metadata
734 * summaries - 'correct' any redundant data
735 * resync - mark array as dirty to trigger a resync.
736 * uuid - set new uuid - only 0.90 or 1.x
737 * name - change the name of the array (where supported)
738 * homehost - change which host this array is tied to.
739 * devicesize - If metadata is at start of device, change recorded
740 * device size to match actual device size
741 * byteorder - swap bytes for 0.90 metadata
742 *
743 * force-one - mark that device as uptodate, not old or failed.
744 * force-array - mark array as clean if it would not otherwise
745 * assemble
746 * assemble - not sure how this is different from force-one...
747 * linear-grow-new - add a new device to a linear array, but don't
748 * change the size: so superblock still matches
749 * linear-grow-update - now change the size of the array.
16715c01
DL
750 * writemostly - set the WriteMostly1 bit in the superblock devflags
751 * readwrite - clear the WriteMostly1 bit in the superblock devflags
199f1a1f
N
752 * no-bitmap - clear any record that a bitmap is present.
753 * bbl - add a bad-block-log if possible
754 * no-bbl - remove and bad-block-log is it is empty.
755 * revert-reshape - If a reshape is in progress, modify metadata so
756 * it will resume going in the opposite direction.
6adfd3af 757 */
68c7d6d7 758 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 759 char *update,
e5eac01f
NB
760 char *devname, int verbose,
761 int uuid_set, char *homehost);
6adfd3af
NB
762
763 /* Create new metadata for new array as described. This could
764 * be a new container, or an array in a pre-existing container.
765 * Also used to zero metadata prior to writing it to invalidate old
766 * metadata.
767 */
3da92f27
NB
768 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
769 unsigned long long size, char *name,
83cd1e97
N
770 char *homehost, int *uuid,
771 unsigned long long data_offset);
6adfd3af
NB
772
773 /* update the metadata to include new device, either at create or
774 * when hot-adding a spare.
775 */
f20c3968 776 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
72ca9bcf
N
777 int fd, char *devname,
778 unsigned long long data_offset);
1a64be56
LM
779 /* update the metadata to delete a device,
780 * when hot-removing.
781 */
782 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
783
784 /* Write metadata to one device when fixing problems or adding
785 * a new device.
786 */
3da92f27 787 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
788
789 /* Write all metadata for this array.
790 */
111d01fc 791 int (*write_init_super)(struct supertype *st);
e2f408a4
N
792 /* Check if metadata read from one device is compatible with an array,
793 * used when assembling an array, or pseudo-assembling was with
794 * "--examine --brief"
795 * If "st" has not yet been loaded the superblock from, "tst" is
796 * moved in, otherwise the superblock in 'st' is compared with
797 * 'tst'.
798 */
64557c33 799 int (*compare_super)(struct supertype *st, struct supertype *tst);
e2f408a4
N
800 /* Load metadata from a single device. If 'devname' is not NULL
801 * print error messages as appropriate */
3da92f27 802 int (*load_super)(struct supertype *st, int fd, char *devname);
e2f408a4
N
803 /* 'fd' is a 'container' md array - load array metadata from the
804 * whole container.
805 */
2b959fbf 806 int (*load_container)(struct supertype *st, int fd, char *devname);
e2f408a4
N
807 /* If 'arg' is a valid name of this metadata type, allocate and
808 * return a 'supertype' for the particular minor version */
82d9eba6 809 struct supertype * (*match_metadata_desc)(char *arg);
e2f408a4
N
810 /* If a device has the given size, and the data_offset has been
811 * requested - work out how much space is available for data.
812 * This involves adjusting for reserved space (e.g. bitmaps)
813 * and for any rounding.
814 * 'mdadm' only calls this for existing arrays where a possible
815 * spare is being added. However some super-handlers call it
816 * internally from validate_geometry when creating an array.
817 */
387fcd59
N
818 __u64 (*avail_size)(struct supertype *st, __u64 size,
819 unsigned long long data_offset);
e2f408a4
N
820 /* This is similar to 'avail_size' in purpose, but is used for
821 * containers for which there is no 'component size' to compare.
822 * This reports that whole-device size which is a minimum
823 */
80e7f8c3 824 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
e2f408a4
N
825 /* Find somewhere to put a bitmap - possibly auto-size it - and
826 * update the metadata to record this. The array may be newly
827 * created, in which case data_size may be updated, or it might
828 * already exist. Metadata handler can know if init_super
829 * has been called, but not write_init_super.
830 */
3da92f27 831 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 832 int delay, int write_behind,
f9c25f1d 833 unsigned long long size, int may_change, int major);
e2f408a4
N
834 /* Seek 'fd' to start of write-intent-bitmap. Must be an
835 * md-native format bitmap
836 */
3da92f27 837 void (*locate_bitmap)(struct supertype *st, int fd);
e2f408a4
N
838 /* if add_internal_bitmap succeeded for existing array, this
839 * writes it out.
840 */
3da92f27 841 int (*write_bitmap)(struct supertype *st, int fd);
e2f408a4 842 /* Free the superblock and any other allocated data */
3da92f27 843 void (*free_super)(struct supertype *st);
78e44928
NB
844
845 /* validate_geometry is called with an st returned by
846 * match_metadata_desc.
e2f408a4 847 * It should check that the geometry described is compatible with
78e44928
NB
848 * the metadata type. It will be called repeatedly as devices
849 * added to validate changing size and new devices. If there are
850 * inter-device dependencies, it should record sufficient details
851 * so these can be validated.
30f58b22 852 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
853 * Return value is:
854 * 1: everything is OK
855 * 0: not OK for some reason - if 'verbose', then error was reported.
856 * -1: st->sb was NULL, 'subdev' is a member of a container of this
e2f408a4 857 * type, but array is not acceptable for some reason
ecbd9e81 858 * message was reported even if verbose is 0.
78e44928 859 */
17f25ca6
NB
860 int (*validate_geometry)(struct supertype *st, int level, int layout,
861 int raiddisks,
c21e737b 862 int *chunk, unsigned long long size,
af4348dd 863 unsigned long long data_offset,
2c514b71
NB
864 char *subdev, unsigned long long *freesize,
865 int verbose);
598f0d58 866
e2f408a4
N
867 /* Return a linked list of 'mdinfo' structures for all arrays
868 * in the container. For non-containers, it is like
869 * getinfo_super with an allocated mdinfo.*/
00bbdbda 870 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
871 /* query the supertype for default geometry */
872 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
873 /* Permit subarray's to be deleted from inactive containers */
874 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 875 /* Permit subarray's to be modified */
a951a4f7 876 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 877 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
878 /* Check if reshape is supported for this external format.
879 * st is obtained from super_by_fd() where st->subarray[0] is
880 * initialized to indicate if reshape is being performed at the
881 * container or subarray level
882 */
016e00f5
AK
883#define APPLY_METADATA_CHANGES 1
884#define ROLLBACK_METADATA_CHANGES 0
885
d04f65f4
N
886 int (*reshape_super)(struct supertype *st,
887 unsigned long long size, int level,
7bc71196 888 int layout, int chunksize, int raid_disks,
41784c88 889 int delta_disks, char *backup, char *dev,
016e00f5 890 int direction,
41784c88 891 int verbose); /* optional */
999b4972
N
892 int (*manage_reshape)( /* optional */
893 int afd, struct mdinfo *sra, struct reshape *reshape,
894 struct supertype *st, unsigned long blocks,
895 int *fds, unsigned long long *offsets,
896 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 897
549e9569 898/* for mdmon */
cba0191b
NB
899 int (*open_new)(struct supertype *c, struct active_array *a,
900 char *inst);
ed9d66aa
NB
901
902 /* Tell the metadata handler the current state of the array.
903 * This covers whether it is known to be consistent (no pending writes)
01f157d7 904 * and how far along a resync is known to have progressed
ed9d66aa
NB
905 * (in a->resync_start).
906 * resync status is really irrelevant if the array is not consistent,
907 * but some metadata (DDF!) have a place to record the distinction.
1011e834 908 * If 'consistent' is '2', then the array can mark it dirty if a
01f157d7
N
909 * resync/recovery/whatever is required, or leave it clean if not.
910 * Return value is 0 dirty (not consistent) and 1 if clean.
911 * it is only really important if consistent is passed in as '2'.
4e5528c6 912 */
01f157d7 913 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
914
915 /* When the state of a device might have changed, we call set_disk to
916 * tell the metadata what the current state is.
917 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
918 * transitions.
919 * set_disk might be called when the state of the particular disk has
920 * not in fact changed.
921 */
8d45d196 922 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
923 void (*sync_metadata)(struct supertype *st);
924 void (*process_update)(struct supertype *st,
925 struct metadata_update *update);
edd8d13c
NB
926 void (*prepare_update)(struct supertype *st,
927 struct metadata_update *update);
549e9569 928
6c3fb95c
NB
929 /* activate_spare will check if the array is degraded and, if it
930 * is, try to find some spare space in the container.
931 * On success, it add appropriate updates (For process_update) to
932 * to the 'updates' list and returns a list of 'mdinfo' identifying
933 * the device, or devices as there might be multiple missing
934 * devices and multiple spares available.
935 */
936 struct mdinfo *(*activate_spare)(struct active_array *a,
937 struct metadata_update **updates);
2cda7640
ML
938 /*
939 * Return statically allocated string that represents metadata specific
940 * controller domain of the disk. The domain is used in disk domain
941 * matching functions. Disks belong to the same domain if the they have
942 * the same domain from mdadm.conf and belong the same metadata domain.
943 * Returning NULL or not providing this handler means that metadata
944 * does not distinguish the differences between disks that belong to
945 * different controllers. They are in the domain specified by
946 * configuration file (mdadm.conf).
947 * In case when the metadata has the notion of domains based on disk
948 * it shall return NULL for disks that do not belong to the controller
949 * the supported domains. Such disks will form another domain and won't
950 * be mixed with supported ones.
951 */
952 const char *(*get_disk_controller_domain)(const char *path);
549e9569 953
ba53ea59
AK
954 /* for external backup area */
955 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
956
f277ce36 957 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 958 int external;
4cce4069 959 const char *name; /* canonical metadata name */
0f22b998 960} *superlist[];
82d9eba6 961
0f22b998
N
962extern struct superswitch super0, super1;
963extern struct superswitch super_imsm, super_ddf;
0592faeb 964extern struct superswitch mbr, gpt;
cdddbdbc 965
edd8d13c
NB
966struct metadata_update {
967 int len;
968 char *buf;
969 void *space; /* allocated space that monitor will use */
cb23f1f4
N
970 void **space_list; /* list of allocated spaces that monitor can
971 * use or that it returned.
972 */
edd8d13c
NB
973 struct metadata_update *next;
974};
975
6adfd3af
NB
976/* A supertype holds a particular collection of metadata.
977 * It identifies the metadata type by the superswitch, and the particular
978 * sub-version of that metadata type.
979 * metadata read in or created is stored in 'sb' and 'info'.
980 * There are also fields used by mdmon to track containers.
981 *
352452c3
N
982 * A supertype may refer to:
983 * Just an array, possibly in a container
984 * A container, not identifying any particular array
985 * Info read from just one device, not yet fully describing the array/container.
986 *
987 *
6adfd3af
NB
988 * A supertype is created by:
989 * super_by_fd
990 * guess_super
991 * dup_super
992 */
82d9eba6
NB
993struct supertype {
994 struct superswitch *ss;
995 int minor_version;
ea329559 996 int max_devs;
4dd2df09 997 char container_devnm[32]; /* devnm of container */
64557c33 998 void *sb;
111d01fc 999 void *info;
afa368f4
N
1000 void *other; /* Hack used to convert v0.90 to v1.0 */
1001 unsigned long long devsize;
23bf42cc 1002 unsigned long long data_offset; /* used by v1.x only */
df3346e6
LM
1003 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
1004 HW/firmware related incompatability to load metadata.
1005 Used when examining metadata to display content of disk
1006 when user has no hw/firmare compatible system.
1007 */
edd8d13c
NB
1008 struct metadata_update *updates;
1009 struct metadata_update **update_tail;
1010
549e9569
NB
1011 /* extra stuff used by mdmon */
1012 struct active_array *arrays;
549e9569 1013 int sock; /* listen to external programs */
4dd2df09
N
1014 char devnm[32]; /* e.g. md0. This appears in metadata_version:
1015 * external:/md0/12
1016 */
549e9569 1017 int devcnt;
68226a80 1018 int retry_soon;
549e9569
NB
1019
1020 struct mdinfo *devs;
1021
82d9eba6 1022};
f9ce90ba 1023
4725bc31 1024extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
1025enum guess_types { guess_any, guess_array, guess_partitions };
1026extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
1027static inline struct supertype *guess_super(int fd) {
1028 return guess_super_type(fd, guess_any);
1029}
3da92f27 1030extern struct supertype *dup_super(struct supertype *st);
beae1dfe 1031extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 1032extern int must_be_container(int fd);
0f0749ad 1033extern int dev_size_from_id(dev_t id, unsigned long long *size);
a7c6e3fb 1034void wait_for(char *dev, int fd);
f9ce90ba 1035
5527fc74
N
1036/*
1037 * Data structures for policy management.
1038 * Each device can have a policy structure that lists
1039 * various name/value pairs each possibly with a metadata associated.
1040 * The policy list is sorted by name/value/metadata
1041 */
1042struct dev_policy {
1043 struct dev_policy *next;
1044 char *name; /* None of these strings are allocated. They are
1045 * all just references to strings which are known
1046 * to exist elsewhere.
1047 * name and metadata can be compared by address equality.
1048 */
1049 const char *metadata;
2cda7640 1050 const char *value;
5527fc74
N
1051};
1052
4e8d9f0a 1053extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
1054
1055/* iterate over the sublist starting at list, having the same
1056 * 'name' as 'list', and matching the given metadata (Where
1057 * NULL matches anything
1058 */
1059#define pol_for_each(item, list, _metadata) \
1060 for (item = list; \
1061 item && item->name == list->name; \
1062 item = item->next) \
1063 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
1064 ; else
1065
1066/*
1067 * policy records read from mdadm are largely just name-value pairs.
1068 * The names are constants, not strdupped
1069 */
1070struct pol_rule {
1071 struct pol_rule *next;
1072 char *type; /* rule_policy or rule_part */
1073 struct rule {
1074 struct rule *next;
1075 char *name;
1076 char *value;
1077 char *dups; /* duplicates of 'value' with a partNN appended */
1078 } *rule;
1079};
1080
1081extern char rule_policy[], rule_part[];
1082extern char rule_path[], rule_type[];
1083extern char type_part[], type_disk[];
1084
1085extern void policyline(char *line, char *type);
4e8d9f0a 1086extern void policy_add(char *type, ...);
5527fc74
N
1087extern void policy_free(void);
1088
73c9c47c 1089extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 1090extern struct dev_policy *disk_policy(struct mdinfo *disk);
4dd2df09 1091extern struct dev_policy *devid_policy(int devid);
5527fc74
N
1092extern void dev_policy_free(struct dev_policy *p);
1093
e78dda3b
N
1094//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
1095extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
1096extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
1097
e3bb5f14
N
1098enum policy_action {
1099 act_default,
1100 act_include,
1101 act_re_add,
d2db3045
N
1102 act_spare, /* This only applies to bare devices */
1103 act_spare_same_slot, /* this allows non-bare devices,
1104 * but only if recent removal */
1105 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
1106 act_err
1107};
1108
1109extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
1110 enum policy_action want);
1111extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
1112 enum policy_action want);
f5f12c84
N
1113
1114struct domainlist {
1115 struct domainlist *next;
2cda7640 1116 const char *dom;
f5f12c84
N
1117};
1118
1119extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
1120 const char *metadata);
1121extern struct domainlist *domain_from_array(struct mdinfo *mdi,
1122 const char *metadata);
4dd2df09 1123extern void domainlist_add_dev(struct domainlist **dom, int devid,
e78dda3b 1124 const char *metadata);
f5f12c84
N
1125extern void domain_free(struct domainlist *dl);
1126extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
1127 const char *metadata);
e78dda3b 1128void domain_add(struct domainlist **domp, char *domain);
f5f12c84 1129
403410eb
PC
1130extern void policy_save_path(char *id_path, struct map_ent *array);
1131extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
1132
838acbc2
NB
1133#if __GNUC__ < 3
1134struct stat64;
1135#endif
1136
45e878bb
NB
1137#define HAVE_NFTW we assume
1138#define HAVE_FTW
1139
2df1f269 1140#ifdef __UCLIBC__
45e878bb 1141# include <features.h>
2df1f269
BRF
1142# ifndef __UCLIBC_HAS_LFS__
1143# define lseek64 lseek
1144# endif
45e878bb
NB
1145# ifndef __UCLIBC_HAS_FTW__
1146# undef HAVE_FTW
1147# undef HAVE_NFTW
1148# endif
1149#endif
1150
1151#ifdef __dietlibc__
1152# undef HAVE_NFTW
1153#endif
1154
e39b673e 1155#if defined(__KLIBC__)
1156# undef HAVE_NFTW
1157# undef HAVE_FTW
1158#endif
1159
45e878bb 1160#ifndef HAVE_NFTW
173fc515 1161# define FTW_PHYS 1
45e878bb
NB
1162# ifndef HAVE_FTW
1163 struct FTW {};
173fc515
NB
1164# endif
1165#endif
1166
45e878bb
NB
1167#ifdef HAVE_FTW
1168# include <ftw.h>
1169#endif
1170
173fc515
NB
1171extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1172
64c4757e 1173extern int Manage_ro(char *devname, int fd, int readonly);
d5a40416 1174extern int Manage_run(char *devname, int fd, struct context *c);
fe7e0e64
N
1175extern int Manage_stop(char *devname, int fd, int quiet,
1176 int will_retry);
64c4757e 1177extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1178 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1179 char *update, int force);
1f48664b 1180extern int autodetect(void);
e5329c37 1181extern int Grow_Add_device(char *devname, int fd, char *newdev);
50f01ba5
N
1182extern int Grow_addbitmap(char *devname, int fd,
1183 struct context *c, struct shape *s);
32754b7d 1184extern int Grow_reshape(char *devname, int fd,
e2e53a2d 1185 struct mddev_dev *devlist,
40c9a66a 1186 unsigned long long data_offset,
32754b7d 1187 struct context *c, struct shape *s);
06b0d786 1188extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1189 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1190extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1191 struct mdinfo *info, char *backup_file,
1192 int freeze_reshape);
1193
3f54bd62
AK
1194extern int restore_backup(struct supertype *st,
1195 struct mdinfo *content,
1196 int working_disks,
1197 int spares,
1198 char *backup_file,
1199 int verbose);
2dddadb0
AK
1200extern int Grow_continue_command(char *devname, int fd,
1201 char *backup_file, int verbose);
64c4757e 1202
7f91af49 1203extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1204 struct mddev_ident *ident,
87f26d14 1205 struct mddev_dev *devlist,
4977146a 1206 struct context *c);
64c4757e 1207
a4e8316a
N
1208extern int Build(char *mddev, struct mddev_dev *devlist,
1209 struct shape *s, struct context *c);
64c4757e 1210
7f91af49 1211extern int Create(struct supertype *st, char *mddev,
171dccc8 1212 char *name, int *uuid,
a655e550 1213 int subdevs, struct mddev_dev *devlist,
99cc42f4 1214 struct shape *s,
40c9a66a
N
1215 struct context *c,
1216 unsigned long long data_offset);
64c4757e 1217
ef898ce6 1218extern int Detail(char *dev, struct context *c);
9eafa1de 1219extern int Detail_Platform(struct superswitch *ss, int scan, int verbose, int export, char *controller_path);
e0d19036 1220extern int Query(char *dev);
6d388a88 1221extern int ExamineBadblocks(char *devname, int brief, struct supertype *forcest);
eec3f887
N
1222extern int Examine(struct mddev_dev *devlist, struct context *c,
1223 struct supertype *forcest);
a655e550 1224extern int Monitor(struct mddev_dev *devlist,
e0d19036 1225 char *mailaddr, char *alert_cmd,
95c50205
N
1226 struct context *c,
1227 int daemonise, int oneshot,
1228 int dosyslog, char *pidfile, int increments,
1229 int share);
64c4757e 1230
ba728be7
N
1231extern int Kill(char *dev, struct supertype *st, int force, int verbose, int noexcl);
1232extern int Kill_subarray(char *dev, char *subarray, int verbose);
fa56eddb 1233extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1234extern int Wait(char *dev);
9f1da824 1235extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1236
11b6d91d
N
1237extern int Incremental(char *devname, struct context *c,
1238 struct supertype *st);
8382f19b 1239extern void RebuildMap(void);
d5a40416 1240extern int IncrementalScan(struct context *c, char *devnm);
950bc344 1241extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1242extern int CreateBitmap(char *filename, int force, char uuid[16],
1243 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1244 unsigned long write_behind,
dcec9ee5
NB
1245 unsigned long long array_size,
1246 int major);
55935d51 1247extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1248extern int Write_rules(char *rule_name);
bf4fb153 1249extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1250extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
74db60b0
N
1251extern int Dump_metadata(char *dev, char *dir, struct context *c,
1252 struct supertype *st);
1253extern int Restore_metadata(char *dev, char *dir, struct context *c,
1254 struct supertype *st, int only);
c82f047c 1255
64c4757e 1256extern int md_get_version(int fd);
e0d19036 1257extern int get_linux_version(void);
bc77ed53 1258extern int mdadm_version(char *version);
51d4261c 1259extern unsigned long long parse_size(char *size);
64c4757e 1260extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1261extern int parse_layout_10(char *layout);
1262extern int parse_layout_faulty(char *layout);
79868890 1263extern long parse_num(char *num);
682c7051
NB
1264extern int check_ext2(int fd, char *name);
1265extern int check_reiser(int fd, char *name);
1266extern int check_raid(int fd, char *name);
53ed6ac3
KW
1267extern int check_partitions(int fd, char *dname,
1268 unsigned long long freesize,
1269 unsigned long long size);
64c4757e 1270
dd0781e5 1271extern int get_mdp_major(void);
8b0dabea 1272extern int dev_open(char *dev, int flags);
4dd2df09
N
1273extern int open_dev(char *devnm);
1274extern int open_dev_flags(char *devnm, int flags);
1275extern int open_dev_excl(char *devnm);
8d80900b 1276extern int is_standard(char *dev, int *nump);
9008ed1c 1277extern int same_dev(char *one, char *two);
9eafa1de 1278extern int compare_paths (char* path1,char* path2);
a7dec3fd 1279extern void enable_fds(int devices);
dd0781e5 1280
f1ae21c4 1281extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1282extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1283extern struct mddev_dev *conf_get_devs(void);
8382f19b 1284extern int conf_test_dev(char *devname);
4e8d9f0a 1285extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1286extern struct createinfo *conf_get_create_info(void);
1287extern void set_conffile(char *file);
1288extern char *conf_get_mailaddr(void);
1289extern char *conf_get_mailfrom(void);
1290extern char *conf_get_program(void);
0ac91628 1291extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1292extern char *conf_line(FILE *file);
98c6faba 1293extern char *conf_word(FILE *file, int allow_key);
7103b9b8
N
1294extern void print_quoted(char *str);
1295extern void print_escape(char *str);
06d2ffc3 1296extern int use_udev(void);
2eba8496 1297extern unsigned long GCD(unsigned long a, unsigned long b);
0ac91628 1298extern int conf_name_is_free(char *name);
7c336758 1299extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1300extern int devname_matches(char *name, char *match);
2244d1a9
N
1301extern struct mddev_ident *conf_match(struct supertype *st,
1302 struct mdinfo *info,
1303 char *devname,
1304 int verbose, int *rvp);
6d11ec6f 1305extern int experimental(void);
0ac91628 1306
e0d19036
NB
1307extern void free_line(char *line);
1308extern int match_oneof(char *devices, char *devname);
e0d19036 1309extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1310extern const int uuid_zero[4];
f277ce36 1311extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1312extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1313extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1314extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1315 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1316extern unsigned long calc_csum(void *super, int bytes);
583315d9 1317extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1318 char *avail);
8453e704 1319extern int enough_fd(int fd);
e0d19036 1320extern int ask(char *mesg);
8fac0577 1321extern unsigned long long get_component_size(int fd);
0430ed48 1322extern void remove_partitions(int fd);
691c6ee1 1323extern int test_partition(int fd);
bfd76b93 1324extern int test_partition_from_id(dev_t id);
577e8448 1325extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1326extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1327 int chunksize, unsigned long long devsize);
edd8d13c
NB
1328extern int flush_metadata_updates(struct supertype *st);
1329extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c 1330extern int assemble_container_content(struct supertype *st, int mdfd,
11b6d91d
N
1331 struct mdinfo *content,
1332 struct context *c,
1333 char *chosen_name);
326727d9
AC
1334extern struct mdinfo *container_choose_spares(struct supertype *st,
1335 unsigned long long min_size,
1336 struct domainlist *domlist,
1337 char *spare_group,
1338 const char *metadata, int get_one);
d52bb542 1339extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1340extern int add_disk(int mdfd, struct supertype *st,
1341 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1342extern int remove_disk(int mdfd, struct supertype *st,
1343 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1344extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1345unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1346
56eb10c0 1347extern char *human_size(long long bytes);
f0ec6710 1348extern char *human_size_brief(long long bytes, int prefix);
e4965ef8 1349extern void print_r10_layout(int layout);
e0d19036 1350
4dd2df09 1351extern char *find_free_devnm(int use_partitions);
63152c1b 1352
e0d19036 1353extern void put_md_name(char *name);
bc17158d 1354extern char *devid2kname(int devid);
4dd2df09
N
1355extern char *devid2devnm(int devid);
1356extern int devnm2devid(char *devnm);
1357extern char *get_md_name(char *devnm);
e0d19036 1358
c913b90e 1359extern char DefaultConfFile[];
b5e64645 1360
69207ff6
N
1361extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1362 char *chosen);
1363/* values for 'trustworthy' */
1364#define LOCAL 1
d1d3482b 1365#define LOCAL_ANY 10
69207ff6
N
1366#define FOREIGN 2
1367#define METADATA 3
6be1d39d 1368extern int open_mddev(char *dev, int report_errors);
a322f70c 1369extern int open_container(int fd);
4dd2df09
N
1370extern int metadata_container_matches(char *metadata, char *devnm);
1371extern int metadata_subdev_matches(char *metadata, char *devnm);
33414a01
DW
1372extern int is_container_member(struct mdstat_ent *ent, char *devname);
1373extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1374extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1375extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1376
4dd2df09
N
1377extern int mdmon_running(char *devnm);
1378extern int mdmon_pid(char *devnm);
40ebbb9c 1379extern int check_env(char *name);
148acb7b 1380extern __u32 random32(void);
4dd2df09 1381extern int start_mdmon(char *devnm);
a931db9e 1382
999b4972
N
1383extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1384 struct supertype *st, unsigned long stripes,
1385 int *fds, unsigned long long *offsets,
1386 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1387void abort_reshape(struct mdinfo *sra);
999b4972 1388
afa368f4
N
1389void *super1_make_v0(struct supertype *st, struct mdinfo *info, mdp_super_t *sb0);
1390
11877f4d 1391extern void fmt_devname(char *name, int num);
4dd2df09
N
1392extern char *stat2devnm(struct stat *st);
1393extern char *fd2devnm(int fd);
a931db9e 1394
a9c15847
N
1395extern int in_initrd(void);
1396
2d762ade 1397#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1398#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1399#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1400 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1401
3c558363
N
1402static inline int is_subarray(char *vers)
1403{
e9dd1598 1404 /* The version string for a 'subarray' (an array in a container)
1011e834 1405 * is
e9dd1598 1406 * /containername/componentname for normal read-write arrays
746a6567
N
1407 * -containername/componentname for arrays which mdmon must not
1408 * reconfigure. They might be read-only
1409 * or might be undergoing reshape etc.
e9dd1598
N
1410 * containername is e.g. md0, md_d1
1411 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1412 */
1413 return (*vers == '/' || *vers == '-');
3c558363
N
1414}
1415
e5408a32
DW
1416static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1417{
1418 return &ent->metadata_version[10+strlen(container)+1];
1419}
1420
2a24d7b6
DW
1421#ifdef DEBUG
1422#define dprintf(fmt, arg...) \
1423 fprintf(stderr, fmt, ##arg)
1424#else
1425#define dprintf(fmt, arg...) \
1426 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1427#endif
1a0ee0ba
DK
1428#include <assert.h>
1429#include <stdarg.h>
1430static inline int xasprintf(char **strp, const char *fmt, ...) {
1431 va_list ap;
1432 int ret;
1433 va_start(ap, fmt);
089485cb 1434 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1435 va_end(ap);
1436 assert(ret >= 0);
1437 return ret;
1438}
2a24d7b6 1439
e7b84f9d
N
1440#define pr_err(fmt ...) fprintf(stderr, Name ": " fmt)
1441#define cont_err(fmt ...) fprintf(stderr, " " fmt)
1442
503975b9
N
1443void *xmalloc(size_t len);
1444void *xrealloc(void *ptr, size_t len);
1445void *xcalloc(size_t num, size_t size);
1446char *xstrdup(const char *str);
1447
b5e64645
NB
1448#define LEVEL_MULTIPATH (-4)
1449#define LEVEL_LINEAR (-1)
1450#define LEVEL_FAULTY (-5)
1451
17f25ca6
NB
1452/* kernel module doesn't know about these */
1453#define LEVEL_CONTAINER (-100)
a322f70c 1454#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1455
b5e64645
NB
1456/* faulty stuff */
1457
1458#define WriteTransient 0
1459#define ReadTransient 1
1460#define WritePersistent 2
1461#define ReadPersistent 3
1462#define WriteAll 4 /* doesn't go to device */
1463#define ReadFixable 5
1464#define Modes 6
1465
1466#define ClearErrors 31
1467#define ClearFaults 30
1468
1469#define AllPersist 100 /* internal use only */
1470#define NoPersist 101
1471
1472#define ModeMask 0x1f
1473#define ModeShift 5
fe6729fa 1474
fe6729fa
NB
1475#ifdef __TINYC__
1476#undef minor
1477#undef major
1478#undef makedev
1479#define minor(x) ((x)&0xff)
1480#define major(x) (((x)>>8)&0xff)
1481#define makedev(M,m) (((M)<<8) | (m))
1482#endif
1483
b640a252 1484/* for raid4/5/6 */
e86c9dd6
NB
1485#define ALGORITHM_LEFT_ASYMMETRIC 0
1486#define ALGORITHM_RIGHT_ASYMMETRIC 1
1487#define ALGORITHM_LEFT_SYMMETRIC 2
1488#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1489
1490/* Define non-rotating (raid4) algorithms. These allow
1491 * conversion of raid4 to raid5.
1492 */
1493#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1494#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1495
1496/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1497 * Firstly, the exact positioning of the parity block is slightly
1498 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1499 * of DDF.
1500 * Secondly, or order of datablocks over which the Q syndrome is computed
1501 * is different.
1502 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1503 * These layouts are from the DDFv1.2 spec.
1504 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1505 * leaves RLQ=3 as 'Vendor Specific'
1506 */
1507
1508#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1509#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1510#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1511
b640a252
N
1512/* For every RAID5 algorithm we define a RAID6 algorithm
1513 * with exactly the same layout for data and parity, and
1514 * with the Q block always on the last device (N-1).
1515 * This allows trivial conversion from RAID5 to RAID6
1516 */
1517#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1518#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1519#define ALGORITHM_LEFT_SYMMETRIC_6 18
1520#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1521#define ALGORITHM_PARITY_0_6 20
1522#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1523
33a6535d
AW
1524/* Define PATH_MAX in case we don't use glibc or standard library does
1525 * not have PATH_MAX defined. Assume max path length is 4K characters.
1526 */
1527#ifndef PATH_MAX
1528#define PATH_MAX 4096
1529#endif
1530
9dad51d4
N
1531#define RESYNC_NONE -1
1532#define RESYNC_DELAYED -2
1533#define RESYNC_PENDING -3
1534#define RESYNC_UNKNOWN -4
a0963a86 1535
480f3566
N
1536/* When using "GET_DISK_INFO" it isn't certain how high
1537 * we need to check. So we impose an absolute limit of
1538 * MAX_DISKS. This needs to be much more than the largest
1539 * number of devices any metadata can support. Currently
1540 * v1.x can support 1920
1541 */
1542#define MAX_DISKS 4096
1543
d04f65f4
N
1544/* Sometimes the 'size' value passed needs to mean "Maximum".
1545 * In those cases with use MAX_SIZE
1546 */
1547#define MAX_SIZE 1
822e393a
N
1548
1549/* We want to use unsigned numbers for sector counts, but need
1550 * a value for 'invalid'. Use '1'.
1551 */
1552#define INVALID_SECTORS 1
72ca9bcf
N
1553/* And another special number needed for --data_offset=variable */
1554#define VARIABLE_OFFSET 3