]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Incremental: improve support for "DEVICE" based restriction in mdadm.conf
[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,
f33a71f1 314 AddSpare,
1c7a808c
N
315 Remove,
316 Fail,
70c55e36
N
317 Replace,
318 With,
1c7a808c
N
319 MiscOpt,
320 WaitOpt,
321 ConfigFile,
322 ChunkSize,
323 WriteMostly,
324 Layout,
325 Auto,
326 Force,
327 SuperMinor,
328 EMail,
329 ProgramOpt,
330 Increment,
331 Fork,
332 Bitmap,
333 RebuildMapOpt,
87f26d14 334 InvalidBackup,
20b60dcd 335 UdevRules,
b76b30e0 336 FreezeReshape,
2dddadb0 337 Continue,
08ca2adf 338 OffRootOpt,
c2ecf5f6 339 Prefer,
f7d3febc 340 KillOpt,
40c9a66a 341 DataOffset,
6d388a88 342 ExamineBB,
74db60b0
N
343 Dump,
344 Restore,
997aed5d
NB
345};
346
f0ec6710
MN
347enum prefix_standard {
348 JEDEC,
349 IEC
350};
351
64c4757e 352/* structures read from config file */
52826846
NB
353/* List of mddevice names and identifiers
354 * Identifiers can be:
355 * uuid=128-hex-uuid
356 * super-minor=decimal-minor-number-from-superblock
357 * devices=comma,separated,list,of,device,names,with,wildcards
358 *
359 * If multiple fields are present, the intersection of all matching
360 * devices is considered
361 */
98c6faba 362#define UnSet (0xfffe)
fa56eddb 363struct mddev_ident {
dd0781e5 364 char *devname;
aba69144 365
dd0781e5 366 int uuid_set;
3fa06e9d 367 int uuid[4];
947fd4dd 368 char name[33];
52826846 369
f21e18ca 370 int super_minor;
52826846 371
dd0781e5 372 char *devices; /* comma separated list of device
52826846
NB
373 * names with wild cards
374 */
dd0781e5 375 int level;
f21e18ca
N
376 int raid_disks;
377 int spare_disks;
82d9eba6 378 struct supertype *st;
dd0781e5
NB
379 int autof; /* 1 for normal, 2 for partitioned */
380 char *spare_group;
7ef02d01 381 char *bitmap_file;
c82f047c 382 int bitmap_fd;
dd0781e5 383
1771a6e2
N
384 char *container; /* /dev/whatever name of container, or
385 * uuid of container. You would expect
386 * this to be the 'devname' or UUID
dbb44303
N
387 * of some other entry.
388 */
389 char *member; /* subarray within a container */
390
fa56eddb 391 struct mddev_ident *next;
b179246f
N
392 union {
393 /* fields needed by different users of this structure */
394 int assembled; /* set when assembly succeeds */
395 };
fa56eddb 396};
64c4757e 397
9e33d556
N
398struct context {
399 int readonly;
400 int runstop;
401 int verbose;
9e33d556
N
402 int brief;
403 int force;
404 char *homehost;
405 int require_homehost;
406 char *prefer;
407 int export;
408 int test;
409 char *subarray;
410 char *update;
411 int scan;
412 int SparcAdjust;
413 int autof;
414 int delay;
415 int freeze_reshape;
416 char *backup_file;
417 int invalid_backup;
418};
419
e705e81b
N
420struct shape {
421 int raiddisks;
422 int sparedisks;
423 int level;
424 int layout;
425 char *layout_str;
426 int chunk;
427 int bitmap_chunk;
428 char *bitmap_file;
429 int assume_clean;
430 int write_behind;
431 unsigned long long size;
432};
433
64c4757e 434/* List of device names - wildcards expanded */
a655e550 435struct mddev_dev {
64c4757e 436 char *devname;
c8e1a230
N
437 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail,
438 * 'A' for re_add.
cd29a5c8
NB
439 * Not set for names read from .config
440 */
b3d31955 441 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
70c55e36 442 int used; /* set when used */
72ca9bcf 443 long long data_offset;
a655e550
N
444 struct mddev_dev *next;
445};
64c4757e 446
682c7051
NB
447typedef struct mapping {
448 char *name;
449 int num;
450} mapping_t;
451
e0d19036
NB
452struct mdstat_ent {
453 char *dev;
4dd2df09 454 char devnm[32];
e0d19036
NB
455 int active;
456 char *level;
457 char *pattern; /* U or up, _ for down */
458 int percent; /* -1 if no resync */
f94c116f 459 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
460 int devcnt;
461 int raid_disks;
549e9569 462 char * metadata_version;
3b57c466
N
463 struct dev_member {
464 char *name;
465 struct dev_member *next;
1011e834 466 } *members;
e0d19036
NB
467 struct mdstat_ent *next;
468};
469
22a88995 470extern struct mdstat_ent *mdstat_read(int hold, int start);
a7a0d8a1 471extern void mdstat_close(void);
e0d19036 472extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 473extern void mdstat_wait(int seconds);
58a4ba2a 474extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
4dd2df09 475extern int mddev_busy(char *devnm);
3b57c466 476extern struct mdstat_ent *mdstat_by_component(char *name);
4dd2df09 477extern struct mdstat_ent *mdstat_by_subdev(char *subdev, char *container);
8382f19b
NB
478
479struct map_ent {
480 struct map_ent *next;
4dd2df09 481 char devnm[32];
1522c538 482 char metadata[20];
8382f19b 483 int uuid[4];
195254b8 484 int bad;
8382f19b
NB
485 char *path;
486};
4dd2df09 487extern int map_update(struct map_ent **mpp, char *devnm, char *metadata,
8382f19b 488 int uuid[4], char *path);
4dd2df09 489extern void map_remove(struct map_ent **map, char *devnm);
8382f19b 490extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
eb2306f8
N
491#ifdef MDASSEMBLE
492static inline struct map_ent *map_by_devnm(struct map_ent **map, char *name)
493{
494 return NULL;
495}
496static inline void map_free(struct map_ent *map)
497{
498}
499#else
4dd2df09 500extern struct map_ent *map_by_devnm(struct map_ent **map, char *devnm);
eb2306f8
N
501extern void map_free(struct map_ent *map);
502#endif
f2e55ecc 503extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
504extern void map_read(struct map_ent **melp);
505extern int map_write(struct map_ent *mel);
4dd2df09 506extern void map_delete(struct map_ent **mapp, char *devnm);
8382f19b 507extern void map_add(struct map_ent **melp,
4dd2df09 508 char *devnm, char *metadata, int uuid[4], char *path);
ad5bc697
N
509extern int map_lock(struct map_ent **melp);
510extern void map_unlock(struct map_ent **melp);
cc700db3 511extern void map_fork(void);
e0d19036 512
e86c9dd6 513/* various details can be requested */
dab4a513
DW
514enum sysfs_read_flags {
515 GET_LEVEL = (1 << 0),
516 GET_LAYOUT = (1 << 1),
517 GET_COMPONENT = (1 << 2),
518 GET_CHUNK = (1 << 3),
519 GET_CACHE = (1 << 4),
520 GET_MISMATCH = (1 << 5),
521 GET_VERSION = (1 << 6),
522 GET_DISKS = (1 << 7),
523 GET_DEGRADED = (1 << 8),
524 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
525 GET_BITMAP_LOCATION = (1 << 10),
526
527 GET_DEVS = (1 << 20), /* gets role, major, minor */
528 GET_OFFSET = (1 << 21),
529 GET_SIZE = (1 << 22),
530 GET_STATE = (1 << 23),
531 GET_ERROR = (1 << 24),
dab4a513 532};
e86c9dd6
NB
533
534/* If fd >= 0, get the array it is open on,
4dd2df09 535 * else use devnm.
e86c9dd6 536 */
4dd2df09
N
537extern int sysfs_open(char *devnm, char *devname, char *attr);
538extern void sysfs_init(struct mdinfo *mdi, int fd, char *devnm);
7e0f6979 539extern void sysfs_free(struct mdinfo *sra);
4dd2df09 540extern struct mdinfo *sysfs_read(int fd, char *devnm, unsigned long options);
1770662b
DW
541extern int sysfs_attr_match(const char *attr, const char *str);
542extern int sysfs_match_word(const char *word, char **list);
7e0f6979 543extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 544 char *name, char *val);
7e0f6979 545extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 546 char *name, unsigned long long val);
012a8641
JS
547extern int sysfs_set_num_signed(struct mdinfo *sra, struct mdinfo *dev,
548 char *name, long long val);
97590376 549extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
550extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
551 char *name);
552extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 553extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 554 char *name, unsigned long long *val);
2eba8496
N
555extern int sysfs_fd_get_two(int fd, unsigned long long *v1, unsigned long long *v2);
556extern int sysfs_get_two(struct mdinfo *sra, struct mdinfo *dev,
557 char *name, unsigned long long *v1, unsigned long long *v2);
7236ee7a 558extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
559extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
560 char *name);
93ecfa01
N
561extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
562 char *name, char *val, int size);
8ed3e5e1 563extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 564extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 565extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 566extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
4dd2df09 567extern int sysfs_unique_holder(char *devnm, long rdev);
bc77ed53 568extern int sysfs_freeze_array(struct mdinfo *sra);
efc67e8e 569extern int sysfs_wait(int fd, int *msec);
c69b251b 570extern int load_sys(char *path, char *buf);
130994cb
AK
571extern int reshape_prepare_fdlist(char *devname,
572 struct mdinfo *sra,
573 int raid_disks,
574 int nrdisks,
575 unsigned long blocks,
576 char *backup_file,
577 int *fdlist,
578 unsigned long long *offsets);
579extern void reshape_free_fdlist(int *fdlist,
580 unsigned long long *offsets,
581 int size);
e6e9d47b
AK
582extern int reshape_open_backup_file(char *backup,
583 int fd,
584 char *devname,
585 long blocks,
586 int *fdlist,
a93f87ee
N
587 unsigned long long *offsets,
588 int restart);
1c009fc2
AK
589extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
590 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
591
592extern int save_stripes(int *source, unsigned long long *offsets,
593 int raid_disks, int chunk_size, int level, int layout,
594 int nwrites, int *dest,
a6288483
N
595 unsigned long long start, unsigned long long length,
596 char *buf);
353632d9
NB
597extern int restore_stripes(int *dest, unsigned long long *offsets,
598 int raid_disks, int chunk_size, int level, int layout,
599 int source, unsigned long long read_offset,
2fcb75ae
AK
600 unsigned long long start, unsigned long long length,
601 char *src_buf);
e86c9dd6 602
52826846
NB
603#ifndef Sendmail
604#define Sendmail "/usr/lib/sendmail -t"
605#endif
606
773135f5
NB
607#define SYSLOG_FACILITY LOG_DAEMON
608
682c7051
NB
609extern char *map_num(mapping_t *map, int num);
610extern int map_name(mapping_t *map, char *name);
b640a252 611extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 612
c2ecf5f6
N
613extern char *map_dev_preferred(int major, int minor, int create,
614 char *prefer);
615static inline char *map_dev(int major, int minor, int create)
616{
617 return map_dev_preferred(major, minor, create, NULL);
618}
64c4757e 619
549e9569 620struct active_array;
2e735d19 621struct metadata_update;
64c4757e 622
80bf9135 623/* 'struct reshape' records the intermediate states of
999b4972
N
624 * a general reshape.
625 * The starting geometry is converted to the 'before' geometry
626 * by at most an atomic level change. They could be the same.
627 * Similarly the 'after' geometry is converted to the final
628 * geometry by at most a level change.
629 * Note that 'before' and 'after' must have the same level.
630 * 'blocks' is the minimum number of sectors for a reshape unit.
631 * This will be a multiple of the stripe size in each of the
632 * 'before' and 'after' geometries.
633 * If 'blocks' is 0, no restriping is necessary.
89ecd3cf
N
634 * 'min_offset_change' is the minimum change to data_offset to
635 * allow the reshape to happen. It is at least the larger of
636 * the old and new chunk sizes, and typically the same as 'blocks'
637 * divided by number of data disks.
999b4972
N
638 */
639struct reshape {
640 int level;
641 int parity; /* number of parity blocks/devices */
642 struct {
643 int layout;
644 int data_disks;
645 } before, after;
646 unsigned long long backup_blocks;
89ecd3cf 647 unsigned long long min_offset_change;
999b4972
N
648 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
649 unsigned long long new_size; /* New size of array in sectors */
650};
651
6adfd3af
NB
652/* A superswitch provides entry point the a metadata handler.
653 *
74db60b0 654 * The superswitch primarily operates on some "metadata" that
6adfd3af
NB
655 * is accessed via the 'supertype'.
656 * This metadata has one of three possible sources.
657 * 1/ It is read from a single device. In this case it may not completely
658 * describe the array or arrays as some information might be on other
659 * devices.
660 * 2/ It is read from all devices in a container. In this case all
661 * information is present.
662 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
663 * be complete once enough ->add_to_super calls have completed.
664 *
665 * When creating an array inside a container, the metadata will be
666 * formed by a combination of 2 and 3. The metadata or the array is read,
667 * then new information is added.
668 *
669 * The metadata must sometimes have a concept of a 'current' array
670 * and a 'current' device.
671 * The 'current' array is set by init_super to be the newly created array,
672 * or is set by super_by_fd when it finds it is looking at an array inside
673 * a container.
674 *
675 * The 'current' device is either the device that the metadata was read from
676 * in case 1, or the last device added by add_to_super in case 3.
677 * Case 2 does not identify a 'current' device.
678 */
f9ce90ba 679extern struct superswitch {
6adfd3af
NB
680
681 /* Used to report details of metadata read from a component
682 * device. ->load_super has been called.
683 */
3da92f27 684 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 685 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 686 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 687 void (*export_examine_super)(struct supertype *st);
6d388a88 688 int (*examine_badblocks)(struct supertype *st, int fd, char *devname);
74db60b0 689 int (*copy_metadata)(struct supertype *st, int from, int to);
6adfd3af
NB
690
691 /* Used to report details of an active array.
692 * ->load_super was possibly given a 'component' string.
693 */
3da92f27 694 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 695 void (*brief_detail_super)(struct supertype *st);
0d726f17 696 void (*export_detail_super)(struct supertype *st);
6adfd3af 697
4cce4069 698 /* Optional: platform hardware / firmware details */
9eafa1de
MN
699 int (*detail_platform)(int verbose, int enumerate_only, char *controller_path);
700 int (*export_detail_platform)(int verbose, char *controller_path);
4cce4069 701
6adfd3af
NB
702 /* Used:
703 * to get uuid to storing in bitmap metadata
704 * and 'reshape' backup-data metadata
705 * To see if a device is being re-added to an array it was part of.
706 */
3da92f27 707 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 708
8592f29d 709 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
710 * the container, or about an individual array within the container.
711 * The determination is made either by:
712 * load_super being given a 'component' string.
713 * validate_geometry determining what to create.
d2ca6449
NB
714 * The info includes both array information and device information.
715 * The particular device should be:
716 * The last device added by add_to_super
717 * The device the metadata was loaded from by load_super
a5d85af7
N
718 * If 'map' is present, then it is an array raid_disks long
719 * (raid_disk must already be set and correct) and it is filled
720 * with 1 for slots that are thought to be active and 0 for slots which
721 * appear to be failed/missing.
95eeceeb 722 * *info is zeroed out before data is added.
6adfd3af 723 */
a5d85af7 724 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 725 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 726 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 727 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 728 */
3da92f27 729 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
730
731 /* Make one of several generic modifications to metadata
732 * prior to assembly (or other times).
733 * sparc2.2 - first bug in early 0.90 metadata
734 * super-minor - change name of 0.90 metadata
735 * summaries - 'correct' any redundant data
736 * resync - mark array as dirty to trigger a resync.
737 * uuid - set new uuid - only 0.90 or 1.x
738 * name - change the name of the array (where supported)
739 * homehost - change which host this array is tied to.
740 * devicesize - If metadata is at start of device, change recorded
741 * device size to match actual device size
742 * byteorder - swap bytes for 0.90 metadata
743 *
744 * force-one - mark that device as uptodate, not old or failed.
745 * force-array - mark array as clean if it would not otherwise
746 * assemble
747 * assemble - not sure how this is different from force-one...
748 * linear-grow-new - add a new device to a linear array, but don't
749 * change the size: so superblock still matches
750 * linear-grow-update - now change the size of the array.
16715c01
DL
751 * writemostly - set the WriteMostly1 bit in the superblock devflags
752 * readwrite - clear the WriteMostly1 bit in the superblock devflags
199f1a1f
N
753 * no-bitmap - clear any record that a bitmap is present.
754 * bbl - add a bad-block-log if possible
755 * no-bbl - remove and bad-block-log is it is empty.
756 * revert-reshape - If a reshape is in progress, modify metadata so
757 * it will resume going in the opposite direction.
6adfd3af 758 */
68c7d6d7 759 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 760 char *update,
e5eac01f
NB
761 char *devname, int verbose,
762 int uuid_set, char *homehost);
6adfd3af
NB
763
764 /* Create new metadata for new array as described. This could
765 * be a new container, or an array in a pre-existing container.
766 * Also used to zero metadata prior to writing it to invalidate old
767 * metadata.
768 */
3da92f27
NB
769 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
770 unsigned long long size, char *name,
83cd1e97
N
771 char *homehost, int *uuid,
772 unsigned long long data_offset);
6adfd3af
NB
773
774 /* update the metadata to include new device, either at create or
775 * when hot-adding a spare.
776 */
f20c3968 777 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
72ca9bcf
N
778 int fd, char *devname,
779 unsigned long long data_offset);
1a64be56
LM
780 /* update the metadata to delete a device,
781 * when hot-removing.
782 */
783 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
784
785 /* Write metadata to one device when fixing problems or adding
786 * a new device.
787 */
3da92f27 788 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
789
790 /* Write all metadata for this array.
791 */
111d01fc 792 int (*write_init_super)(struct supertype *st);
e2f408a4
N
793 /* Check if metadata read from one device is compatible with an array,
794 * used when assembling an array, or pseudo-assembling was with
795 * "--examine --brief"
796 * If "st" has not yet been loaded the superblock from, "tst" is
797 * moved in, otherwise the superblock in 'st' is compared with
798 * 'tst'.
799 */
64557c33 800 int (*compare_super)(struct supertype *st, struct supertype *tst);
e2f408a4
N
801 /* Load metadata from a single device. If 'devname' is not NULL
802 * print error messages as appropriate */
3da92f27 803 int (*load_super)(struct supertype *st, int fd, char *devname);
e2f408a4
N
804 /* 'fd' is a 'container' md array - load array metadata from the
805 * whole container.
806 */
2b959fbf 807 int (*load_container)(struct supertype *st, int fd, char *devname);
e2f408a4
N
808 /* If 'arg' is a valid name of this metadata type, allocate and
809 * return a 'supertype' for the particular minor version */
82d9eba6 810 struct supertype * (*match_metadata_desc)(char *arg);
e2f408a4
N
811 /* If a device has the given size, and the data_offset has been
812 * requested - work out how much space is available for data.
813 * This involves adjusting for reserved space (e.g. bitmaps)
814 * and for any rounding.
815 * 'mdadm' only calls this for existing arrays where a possible
816 * spare is being added. However some super-handlers call it
817 * internally from validate_geometry when creating an array.
818 */
387fcd59
N
819 __u64 (*avail_size)(struct supertype *st, __u64 size,
820 unsigned long long data_offset);
e2f408a4
N
821 /* This is similar to 'avail_size' in purpose, but is used for
822 * containers for which there is no 'component size' to compare.
823 * This reports that whole-device size which is a minimum
824 */
80e7f8c3 825 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
e2f408a4
N
826 /* Find somewhere to put a bitmap - possibly auto-size it - and
827 * update the metadata to record this. The array may be newly
828 * created, in which case data_size may be updated, or it might
829 * already exist. Metadata handler can know if init_super
830 * has been called, but not write_init_super.
831 */
3da92f27 832 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 833 int delay, int write_behind,
f9c25f1d 834 unsigned long long size, int may_change, int major);
e2f408a4
N
835 /* Seek 'fd' to start of write-intent-bitmap. Must be an
836 * md-native format bitmap
837 */
3da92f27 838 void (*locate_bitmap)(struct supertype *st, int fd);
e2f408a4
N
839 /* if add_internal_bitmap succeeded for existing array, this
840 * writes it out.
841 */
3da92f27 842 int (*write_bitmap)(struct supertype *st, int fd);
e2f408a4 843 /* Free the superblock and any other allocated data */
3da92f27 844 void (*free_super)(struct supertype *st);
78e44928
NB
845
846 /* validate_geometry is called with an st returned by
847 * match_metadata_desc.
e2f408a4 848 * It should check that the geometry described is compatible with
78e44928
NB
849 * the metadata type. It will be called repeatedly as devices
850 * added to validate changing size and new devices. If there are
851 * inter-device dependencies, it should record sufficient details
852 * so these can be validated.
30f58b22 853 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
854 * Return value is:
855 * 1: everything is OK
856 * 0: not OK for some reason - if 'verbose', then error was reported.
857 * -1: st->sb was NULL, 'subdev' is a member of a container of this
e2f408a4 858 * type, but array is not acceptable for some reason
ecbd9e81 859 * message was reported even if verbose is 0.
78e44928 860 */
17f25ca6
NB
861 int (*validate_geometry)(struct supertype *st, int level, int layout,
862 int raiddisks,
c21e737b 863 int *chunk, unsigned long long size,
af4348dd 864 unsigned long long data_offset,
2c514b71
NB
865 char *subdev, unsigned long long *freesize,
866 int verbose);
598f0d58 867
e2f408a4
N
868 /* Return a linked list of 'mdinfo' structures for all arrays
869 * in the container. For non-containers, it is like
870 * getinfo_super with an allocated mdinfo.*/
00bbdbda 871 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
872 /* query the supertype for default geometry */
873 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
874 /* Permit subarray's to be deleted from inactive containers */
875 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 876 /* Permit subarray's to be modified */
a951a4f7 877 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 878 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
879 /* Check if reshape is supported for this external format.
880 * st is obtained from super_by_fd() where st->subarray[0] is
881 * initialized to indicate if reshape is being performed at the
882 * container or subarray level
883 */
016e00f5
AK
884#define APPLY_METADATA_CHANGES 1
885#define ROLLBACK_METADATA_CHANGES 0
886
d04f65f4
N
887 int (*reshape_super)(struct supertype *st,
888 unsigned long long size, int level,
7bc71196 889 int layout, int chunksize, int raid_disks,
41784c88 890 int delta_disks, char *backup, char *dev,
016e00f5 891 int direction,
41784c88 892 int verbose); /* optional */
999b4972
N
893 int (*manage_reshape)( /* optional */
894 int afd, struct mdinfo *sra, struct reshape *reshape,
895 struct supertype *st, unsigned long blocks,
896 int *fds, unsigned long long *offsets,
897 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 898
549e9569 899/* for mdmon */
cba0191b
NB
900 int (*open_new)(struct supertype *c, struct active_array *a,
901 char *inst);
ed9d66aa
NB
902
903 /* Tell the metadata handler the current state of the array.
904 * This covers whether it is known to be consistent (no pending writes)
01f157d7 905 * and how far along a resync is known to have progressed
ed9d66aa
NB
906 * (in a->resync_start).
907 * resync status is really irrelevant if the array is not consistent,
908 * but some metadata (DDF!) have a place to record the distinction.
1011e834 909 * If 'consistent' is '2', then the array can mark it dirty if a
01f157d7
N
910 * resync/recovery/whatever is required, or leave it clean if not.
911 * Return value is 0 dirty (not consistent) and 1 if clean.
912 * it is only really important if consistent is passed in as '2'.
4e5528c6 913 */
01f157d7 914 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
915
916 /* When the state of a device might have changed, we call set_disk to
917 * tell the metadata what the current state is.
918 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
919 * transitions.
920 * set_disk might be called when the state of the particular disk has
921 * not in fact changed.
922 */
8d45d196 923 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
924 void (*sync_metadata)(struct supertype *st);
925 void (*process_update)(struct supertype *st,
926 struct metadata_update *update);
edd8d13c
NB
927 void (*prepare_update)(struct supertype *st,
928 struct metadata_update *update);
549e9569 929
6c3fb95c
NB
930 /* activate_spare will check if the array is degraded and, if it
931 * is, try to find some spare space in the container.
932 * On success, it add appropriate updates (For process_update) to
933 * to the 'updates' list and returns a list of 'mdinfo' identifying
934 * the device, or devices as there might be multiple missing
935 * devices and multiple spares available.
936 */
937 struct mdinfo *(*activate_spare)(struct active_array *a,
938 struct metadata_update **updates);
2cda7640
ML
939 /*
940 * Return statically allocated string that represents metadata specific
941 * controller domain of the disk. The domain is used in disk domain
942 * matching functions. Disks belong to the same domain if the they have
943 * the same domain from mdadm.conf and belong the same metadata domain.
944 * Returning NULL or not providing this handler means that metadata
945 * does not distinguish the differences between disks that belong to
946 * different controllers. They are in the domain specified by
947 * configuration file (mdadm.conf).
948 * In case when the metadata has the notion of domains based on disk
949 * it shall return NULL for disks that do not belong to the controller
950 * the supported domains. Such disks will form another domain and won't
951 * be mixed with supported ones.
952 */
953 const char *(*get_disk_controller_domain)(const char *path);
549e9569 954
ba53ea59
AK
955 /* for external backup area */
956 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
957
f277ce36 958 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 959 int external;
4cce4069 960 const char *name; /* canonical metadata name */
0f22b998 961} *superlist[];
82d9eba6 962
0f22b998
N
963extern struct superswitch super0, super1;
964extern struct superswitch super_imsm, super_ddf;
0592faeb 965extern struct superswitch mbr, gpt;
cdddbdbc 966
edd8d13c
NB
967struct metadata_update {
968 int len;
969 char *buf;
970 void *space; /* allocated space that monitor will use */
cb23f1f4
N
971 void **space_list; /* list of allocated spaces that monitor can
972 * use or that it returned.
973 */
edd8d13c
NB
974 struct metadata_update *next;
975};
976
6adfd3af
NB
977/* A supertype holds a particular collection of metadata.
978 * It identifies the metadata type by the superswitch, and the particular
979 * sub-version of that metadata type.
980 * metadata read in or created is stored in 'sb' and 'info'.
981 * There are also fields used by mdmon to track containers.
982 *
352452c3
N
983 * A supertype may refer to:
984 * Just an array, possibly in a container
985 * A container, not identifying any particular array
986 * Info read from just one device, not yet fully describing the array/container.
987 *
988 *
6adfd3af
NB
989 * A supertype is created by:
990 * super_by_fd
991 * guess_super
992 * dup_super
993 */
82d9eba6
NB
994struct supertype {
995 struct superswitch *ss;
996 int minor_version;
ea329559 997 int max_devs;
4dd2df09 998 char container_devnm[32]; /* devnm of container */
64557c33 999 void *sb;
111d01fc 1000 void *info;
afa368f4
N
1001 void *other; /* Hack used to convert v0.90 to v1.0 */
1002 unsigned long long devsize;
23bf42cc 1003 unsigned long long data_offset; /* used by v1.x only */
df3346e6
LM
1004 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
1005 HW/firmware related incompatability to load metadata.
1006 Used when examining metadata to display content of disk
1007 when user has no hw/firmare compatible system.
1008 */
edd8d13c
NB
1009 struct metadata_update *updates;
1010 struct metadata_update **update_tail;
1011
549e9569
NB
1012 /* extra stuff used by mdmon */
1013 struct active_array *arrays;
549e9569 1014 int sock; /* listen to external programs */
4dd2df09
N
1015 char devnm[32]; /* e.g. md0. This appears in metadata_version:
1016 * external:/md0/12
1017 */
549e9569 1018 int devcnt;
68226a80 1019 int retry_soon;
549e9569
NB
1020
1021 struct mdinfo *devs;
1022
82d9eba6 1023};
f9ce90ba 1024
4725bc31 1025extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
1026enum guess_types { guess_any, guess_array, guess_partitions };
1027extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
1028static inline struct supertype *guess_super(int fd) {
1029 return guess_super_type(fd, guess_any);
1030}
3da92f27 1031extern struct supertype *dup_super(struct supertype *st);
beae1dfe 1032extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 1033extern int must_be_container(int fd);
0f0749ad 1034extern int dev_size_from_id(dev_t id, unsigned long long *size);
a7c6e3fb 1035void wait_for(char *dev, int fd);
f9ce90ba 1036
5527fc74
N
1037/*
1038 * Data structures for policy management.
1039 * Each device can have a policy structure that lists
1040 * various name/value pairs each possibly with a metadata associated.
1041 * The policy list is sorted by name/value/metadata
1042 */
1043struct dev_policy {
1044 struct dev_policy *next;
1045 char *name; /* None of these strings are allocated. They are
1046 * all just references to strings which are known
1047 * to exist elsewhere.
1048 * name and metadata can be compared by address equality.
1049 */
1050 const char *metadata;
2cda7640 1051 const char *value;
5527fc74
N
1052};
1053
4e8d9f0a 1054extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
1055
1056/* iterate over the sublist starting at list, having the same
1057 * 'name' as 'list', and matching the given metadata (Where
1058 * NULL matches anything
1059 */
1060#define pol_for_each(item, list, _metadata) \
1061 for (item = list; \
1062 item && item->name == list->name; \
1063 item = item->next) \
1064 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
1065 ; else
1066
1067/*
1068 * policy records read from mdadm are largely just name-value pairs.
1069 * The names are constants, not strdupped
1070 */
1071struct pol_rule {
1072 struct pol_rule *next;
1073 char *type; /* rule_policy or rule_part */
1074 struct rule {
1075 struct rule *next;
1076 char *name;
1077 char *value;
1078 char *dups; /* duplicates of 'value' with a partNN appended */
1079 } *rule;
1080};
1081
1082extern char rule_policy[], rule_part[];
1083extern char rule_path[], rule_type[];
1084extern char type_part[], type_disk[];
1085
1086extern void policyline(char *line, char *type);
4e8d9f0a 1087extern void policy_add(char *type, ...);
5527fc74
N
1088extern void policy_free(void);
1089
73c9c47c 1090extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 1091extern struct dev_policy *disk_policy(struct mdinfo *disk);
4dd2df09 1092extern struct dev_policy *devid_policy(int devid);
5527fc74
N
1093extern void dev_policy_free(struct dev_policy *p);
1094
e78dda3b
N
1095//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
1096extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
1097extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
1098
e3bb5f14
N
1099enum policy_action {
1100 act_default,
1101 act_include,
1102 act_re_add,
d2db3045
N
1103 act_spare, /* This only applies to bare devices */
1104 act_spare_same_slot, /* this allows non-bare devices,
1105 * but only if recent removal */
1106 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
1107 act_err
1108};
1109
1110extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
1111 enum policy_action want);
1112extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
1113 enum policy_action want);
f5f12c84
N
1114
1115struct domainlist {
1116 struct domainlist *next;
2cda7640 1117 const char *dom;
f5f12c84
N
1118};
1119
1120extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
1121 const char *metadata);
1122extern struct domainlist *domain_from_array(struct mdinfo *mdi,
1123 const char *metadata);
4dd2df09 1124extern void domainlist_add_dev(struct domainlist **dom, int devid,
e78dda3b 1125 const char *metadata);
f5f12c84
N
1126extern void domain_free(struct domainlist *dl);
1127extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
1128 const char *metadata);
e78dda3b 1129void domain_add(struct domainlist **domp, char *domain);
f5f12c84 1130
403410eb
PC
1131extern void policy_save_path(char *id_path, struct map_ent *array);
1132extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
1133
838acbc2
NB
1134#if __GNUC__ < 3
1135struct stat64;
1136#endif
1137
45e878bb
NB
1138#define HAVE_NFTW we assume
1139#define HAVE_FTW
1140
2df1f269 1141#ifdef __UCLIBC__
45e878bb 1142# include <features.h>
2df1f269
BRF
1143# ifndef __UCLIBC_HAS_LFS__
1144# define lseek64 lseek
1145# endif
45e878bb
NB
1146# ifndef __UCLIBC_HAS_FTW__
1147# undef HAVE_FTW
1148# undef HAVE_NFTW
1149# endif
1150#endif
1151
1152#ifdef __dietlibc__
1153# undef HAVE_NFTW
1154#endif
1155
e39b673e 1156#if defined(__KLIBC__)
1157# undef HAVE_NFTW
1158# undef HAVE_FTW
1159#endif
1160
45e878bb 1161#ifndef HAVE_NFTW
173fc515 1162# define FTW_PHYS 1
45e878bb
NB
1163# ifndef HAVE_FTW
1164 struct FTW {};
173fc515
NB
1165# endif
1166#endif
1167
45e878bb
NB
1168#ifdef HAVE_FTW
1169# include <ftw.h>
1170#endif
1171
173fc515
NB
1172extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1173
64c4757e 1174extern int Manage_ro(char *devname, int fd, int readonly);
d5a40416 1175extern int Manage_run(char *devname, int fd, struct context *c);
fe7e0e64
N
1176extern int Manage_stop(char *devname, int fd, int quiet,
1177 int will_retry);
64c4757e 1178extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1179 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1180 char *update, int force);
1f48664b 1181extern int autodetect(void);
e5329c37 1182extern int Grow_Add_device(char *devname, int fd, char *newdev);
50f01ba5
N
1183extern int Grow_addbitmap(char *devname, int fd,
1184 struct context *c, struct shape *s);
32754b7d 1185extern int Grow_reshape(char *devname, int fd,
e2e53a2d 1186 struct mddev_dev *devlist,
40c9a66a 1187 unsigned long long data_offset,
32754b7d 1188 struct context *c, struct shape *s);
06b0d786 1189extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1190 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1191extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1192 struct mdinfo *info, char *backup_file,
1193 int freeze_reshape);
1194
3f54bd62
AK
1195extern int restore_backup(struct supertype *st,
1196 struct mdinfo *content,
1197 int working_disks,
1198 int spares,
1199 char *backup_file,
1200 int verbose);
2dddadb0
AK
1201extern int Grow_continue_command(char *devname, int fd,
1202 char *backup_file, int verbose);
64c4757e 1203
7f91af49 1204extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1205 struct mddev_ident *ident,
87f26d14 1206 struct mddev_dev *devlist,
4977146a 1207 struct context *c);
64c4757e 1208
a4e8316a
N
1209extern int Build(char *mddev, struct mddev_dev *devlist,
1210 struct shape *s, struct context *c);
64c4757e 1211
7f91af49 1212extern int Create(struct supertype *st, char *mddev,
171dccc8 1213 char *name, int *uuid,
a655e550 1214 int subdevs, struct mddev_dev *devlist,
99cc42f4 1215 struct shape *s,
40c9a66a
N
1216 struct context *c,
1217 unsigned long long data_offset);
64c4757e 1218
ef898ce6 1219extern int Detail(char *dev, struct context *c);
9eafa1de 1220extern int Detail_Platform(struct superswitch *ss, int scan, int verbose, int export, char *controller_path);
e0d19036 1221extern int Query(char *dev);
6d388a88 1222extern int ExamineBadblocks(char *devname, int brief, struct supertype *forcest);
eec3f887
N
1223extern int Examine(struct mddev_dev *devlist, struct context *c,
1224 struct supertype *forcest);
a655e550 1225extern int Monitor(struct mddev_dev *devlist,
e0d19036 1226 char *mailaddr, char *alert_cmd,
95c50205
N
1227 struct context *c,
1228 int daemonise, int oneshot,
1229 int dosyslog, char *pidfile, int increments,
1230 int share);
64c4757e 1231
ba728be7
N
1232extern int Kill(char *dev, struct supertype *st, int force, int verbose, int noexcl);
1233extern int Kill_subarray(char *dev, char *subarray, int verbose);
fa56eddb 1234extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1235extern int Wait(char *dev);
9f1da824 1236extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1237
b11fe74d 1238extern int Incremental(struct mddev_dev *devlist, struct context *c,
11b6d91d 1239 struct supertype *st);
8382f19b 1240extern void RebuildMap(void);
d5a40416 1241extern int IncrementalScan(struct context *c, char *devnm);
950bc344 1242extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1243extern int CreateBitmap(char *filename, int force, char uuid[16],
1244 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1245 unsigned long write_behind,
dcec9ee5
NB
1246 unsigned long long array_size,
1247 int major);
55935d51 1248extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1249extern int Write_rules(char *rule_name);
bf4fb153 1250extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1251extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
74db60b0
N
1252extern int Dump_metadata(char *dev, char *dir, struct context *c,
1253 struct supertype *st);
1254extern int Restore_metadata(char *dev, char *dir, struct context *c,
1255 struct supertype *st, int only);
c82f047c 1256
64c4757e 1257extern int md_get_version(int fd);
e0d19036 1258extern int get_linux_version(void);
bc77ed53 1259extern int mdadm_version(char *version);
51d4261c 1260extern unsigned long long parse_size(char *size);
64c4757e 1261extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1262extern int parse_layout_10(char *layout);
1263extern int parse_layout_faulty(char *layout);
79868890 1264extern long parse_num(char *num);
682c7051
NB
1265extern int check_ext2(int fd, char *name);
1266extern int check_reiser(int fd, char *name);
1267extern int check_raid(int fd, char *name);
53ed6ac3
KW
1268extern int check_partitions(int fd, char *dname,
1269 unsigned long long freesize,
1270 unsigned long long size);
64c4757e 1271
dd0781e5 1272extern int get_mdp_major(void);
8b0dabea 1273extern int dev_open(char *dev, int flags);
4dd2df09
N
1274extern int open_dev(char *devnm);
1275extern int open_dev_flags(char *devnm, int flags);
1276extern int open_dev_excl(char *devnm);
8d80900b 1277extern int is_standard(char *dev, int *nump);
9008ed1c 1278extern int same_dev(char *one, char *two);
9eafa1de 1279extern int compare_paths (char* path1,char* path2);
a7dec3fd 1280extern void enable_fds(int devices);
dd0781e5 1281
f1ae21c4 1282extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1283extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1284extern struct mddev_dev *conf_get_devs(void);
8382f19b 1285extern int conf_test_dev(char *devname);
4e8d9f0a 1286extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1287extern struct createinfo *conf_get_create_info(void);
1288extern void set_conffile(char *file);
1289extern char *conf_get_mailaddr(void);
1290extern char *conf_get_mailfrom(void);
1291extern char *conf_get_program(void);
0ac91628 1292extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1293extern char *conf_line(FILE *file);
98c6faba 1294extern char *conf_word(FILE *file, int allow_key);
7103b9b8
N
1295extern void print_quoted(char *str);
1296extern void print_escape(char *str);
06d2ffc3 1297extern int use_udev(void);
2eba8496 1298extern unsigned long GCD(unsigned long a, unsigned long b);
0ac91628 1299extern int conf_name_is_free(char *name);
7c336758 1300extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1301extern int devname_matches(char *name, char *match);
2244d1a9
N
1302extern struct mddev_ident *conf_match(struct supertype *st,
1303 struct mdinfo *info,
1304 char *devname,
1305 int verbose, int *rvp);
6d11ec6f 1306extern int experimental(void);
0ac91628 1307
e0d19036
NB
1308extern void free_line(char *line);
1309extern int match_oneof(char *devices, char *devname);
e0d19036 1310extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1311extern const int uuid_zero[4];
f277ce36 1312extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1313extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1314extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1315extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1316 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1317extern unsigned long calc_csum(void *super, int bytes);
583315d9 1318extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1319 char *avail);
8453e704 1320extern int enough_fd(int fd);
e0d19036 1321extern int ask(char *mesg);
8fac0577 1322extern unsigned long long get_component_size(int fd);
0430ed48 1323extern void remove_partitions(int fd);
691c6ee1 1324extern int test_partition(int fd);
bfd76b93 1325extern int test_partition_from_id(dev_t id);
577e8448 1326extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1327extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1328 int chunksize, unsigned long long devsize);
edd8d13c
NB
1329extern int flush_metadata_updates(struct supertype *st);
1330extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c 1331extern int assemble_container_content(struct supertype *st, int mdfd,
11b6d91d
N
1332 struct mdinfo *content,
1333 struct context *c,
9ca39acb
N
1334 char *chosen_name, int *result);
1335#define INCR_NO 1
1336#define INCR_UNSAFE 2
1337#define INCR_ALREADY 4
1338#define INCR_YES 8
326727d9
AC
1339extern struct mdinfo *container_choose_spares(struct supertype *st,
1340 unsigned long long min_size,
1341 struct domainlist *domlist,
1342 char *spare_group,
1343 const char *metadata, int get_one);
d52bb542 1344extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1345extern int add_disk(int mdfd, struct supertype *st,
1346 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1347extern int remove_disk(int mdfd, struct supertype *st,
1348 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1349extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1350unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1351
56eb10c0 1352extern char *human_size(long long bytes);
f0ec6710 1353extern char *human_size_brief(long long bytes, int prefix);
e4965ef8 1354extern void print_r10_layout(int layout);
e0d19036 1355
4dd2df09 1356extern char *find_free_devnm(int use_partitions);
63152c1b 1357
e0d19036 1358extern void put_md_name(char *name);
bc17158d 1359extern char *devid2kname(int devid);
4dd2df09
N
1360extern char *devid2devnm(int devid);
1361extern int devnm2devid(char *devnm);
1362extern char *get_md_name(char *devnm);
e0d19036 1363
c913b90e 1364extern char DefaultConfFile[];
b5e64645 1365
69207ff6
N
1366extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1367 char *chosen);
1368/* values for 'trustworthy' */
1369#define LOCAL 1
d1d3482b 1370#define LOCAL_ANY 10
69207ff6
N
1371#define FOREIGN 2
1372#define METADATA 3
6be1d39d 1373extern int open_mddev(char *dev, int report_errors);
a322f70c 1374extern int open_container(int fd);
4dd2df09
N
1375extern int metadata_container_matches(char *metadata, char *devnm);
1376extern int metadata_subdev_matches(char *metadata, char *devnm);
33414a01
DW
1377extern int is_container_member(struct mdstat_ent *ent, char *devname);
1378extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1379extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1380extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1381
4dd2df09
N
1382extern int mdmon_running(char *devnm);
1383extern int mdmon_pid(char *devnm);
40ebbb9c 1384extern int check_env(char *name);
148acb7b 1385extern __u32 random32(void);
4dd2df09 1386extern int start_mdmon(char *devnm);
a931db9e 1387
999b4972
N
1388extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1389 struct supertype *st, unsigned long stripes,
1390 int *fds, unsigned long long *offsets,
1391 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1392void abort_reshape(struct mdinfo *sra);
999b4972 1393
afa368f4
N
1394void *super1_make_v0(struct supertype *st, struct mdinfo *info, mdp_super_t *sb0);
1395
11877f4d 1396extern void fmt_devname(char *name, int num);
4dd2df09
N
1397extern char *stat2devnm(struct stat *st);
1398extern char *fd2devnm(int fd);
a931db9e 1399
a9c15847
N
1400extern int in_initrd(void);
1401
2d762ade 1402#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1403#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1404#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1405 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1406
3c558363
N
1407static inline int is_subarray(char *vers)
1408{
e9dd1598 1409 /* The version string for a 'subarray' (an array in a container)
1011e834 1410 * is
e9dd1598 1411 * /containername/componentname for normal read-write arrays
746a6567
N
1412 * -containername/componentname for arrays which mdmon must not
1413 * reconfigure. They might be read-only
1414 * or might be undergoing reshape etc.
e9dd1598
N
1415 * containername is e.g. md0, md_d1
1416 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1417 */
1418 return (*vers == '/' || *vers == '-');
3c558363
N
1419}
1420
e5408a32
DW
1421static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1422{
1423 return &ent->metadata_version[10+strlen(container)+1];
1424}
1425
2a24d7b6
DW
1426#ifdef DEBUG
1427#define dprintf(fmt, arg...) \
1428 fprintf(stderr, fmt, ##arg)
1429#else
1430#define dprintf(fmt, arg...) \
1431 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1432#endif
1a0ee0ba
DK
1433#include <assert.h>
1434#include <stdarg.h>
1435static inline int xasprintf(char **strp, const char *fmt, ...) {
1436 va_list ap;
1437 int ret;
1438 va_start(ap, fmt);
089485cb 1439 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1440 va_end(ap);
1441 assert(ret >= 0);
1442 return ret;
1443}
2a24d7b6 1444
e7b84f9d
N
1445#define pr_err(fmt ...) fprintf(stderr, Name ": " fmt)
1446#define cont_err(fmt ...) fprintf(stderr, " " fmt)
1447
503975b9
N
1448void *xmalloc(size_t len);
1449void *xrealloc(void *ptr, size_t len);
1450void *xcalloc(size_t num, size_t size);
1451char *xstrdup(const char *str);
1452
b5e64645
NB
1453#define LEVEL_MULTIPATH (-4)
1454#define LEVEL_LINEAR (-1)
1455#define LEVEL_FAULTY (-5)
1456
17f25ca6
NB
1457/* kernel module doesn't know about these */
1458#define LEVEL_CONTAINER (-100)
a322f70c 1459#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1460
b5e64645
NB
1461/* faulty stuff */
1462
1463#define WriteTransient 0
1464#define ReadTransient 1
1465#define WritePersistent 2
1466#define ReadPersistent 3
1467#define WriteAll 4 /* doesn't go to device */
1468#define ReadFixable 5
1469#define Modes 6
1470
1471#define ClearErrors 31
1472#define ClearFaults 30
1473
1474#define AllPersist 100 /* internal use only */
1475#define NoPersist 101
1476
1477#define ModeMask 0x1f
1478#define ModeShift 5
fe6729fa 1479
fe6729fa
NB
1480#ifdef __TINYC__
1481#undef minor
1482#undef major
1483#undef makedev
1484#define minor(x) ((x)&0xff)
1485#define major(x) (((x)>>8)&0xff)
1486#define makedev(M,m) (((M)<<8) | (m))
1487#endif
1488
b640a252 1489/* for raid4/5/6 */
e86c9dd6
NB
1490#define ALGORITHM_LEFT_ASYMMETRIC 0
1491#define ALGORITHM_RIGHT_ASYMMETRIC 1
1492#define ALGORITHM_LEFT_SYMMETRIC 2
1493#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1494
1495/* Define non-rotating (raid4) algorithms. These allow
1496 * conversion of raid4 to raid5.
1497 */
1498#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1499#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1500
1501/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1502 * Firstly, the exact positioning of the parity block is slightly
1503 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1504 * of DDF.
1505 * Secondly, or order of datablocks over which the Q syndrome is computed
1506 * is different.
1507 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1508 * These layouts are from the DDFv1.2 spec.
1509 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1510 * leaves RLQ=3 as 'Vendor Specific'
1511 */
1512
1513#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1514#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1515#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1516
b640a252
N
1517/* For every RAID5 algorithm we define a RAID6 algorithm
1518 * with exactly the same layout for data and parity, and
1519 * with the Q block always on the last device (N-1).
1520 * This allows trivial conversion from RAID5 to RAID6
1521 */
1522#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1523#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1524#define ALGORITHM_LEFT_SYMMETRIC_6 18
1525#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1526#define ALGORITHM_PARITY_0_6 20
1527#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1528
33a6535d
AW
1529/* Define PATH_MAX in case we don't use glibc or standard library does
1530 * not have PATH_MAX defined. Assume max path length is 4K characters.
1531 */
1532#ifndef PATH_MAX
1533#define PATH_MAX 4096
1534#endif
1535
9dad51d4
N
1536#define RESYNC_NONE -1
1537#define RESYNC_DELAYED -2
1538#define RESYNC_PENDING -3
1539#define RESYNC_UNKNOWN -4
a0963a86 1540
480f3566
N
1541/* When using "GET_DISK_INFO" it isn't certain how high
1542 * we need to check. So we impose an absolute limit of
1543 * MAX_DISKS. This needs to be much more than the largest
1544 * number of devices any metadata can support. Currently
1545 * v1.x can support 1920
1546 */
1547#define MAX_DISKS 4096
1548
d04f65f4
N
1549/* Sometimes the 'size' value passed needs to mean "Maximum".
1550 * In those cases with use MAX_SIZE
1551 */
1552#define MAX_SIZE 1
822e393a
N
1553
1554/* We want to use unsigned numbers for sector counts, but need
1555 * a value for 'invalid'. Use '1'.
1556 */
1557#define INVALID_SECTORS 1
72ca9bcf
N
1558/* And another special number needed for --data_offset=variable */
1559#define VARIABLE_OFFSET 3