]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Grow/raid10: support reducing the devices in a RAID10.
[thirdparty/mdadm.git] / mdadm.h
CommitLineData
64c4757e 1/*
9a9dab36 2 * mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e 3 *
e736b623 4 * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
64c4757e
NB
5 *
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 * Author: Neil Brown
e736b623 22 * Email: <neilb@suse.de>
64c4757e
NB
23 */
24
4ccdb956 25#define _GNU_SOURCE
c2c9bb6f 26#define _FILE_OFFSET_BITS 64
64c4757e 27#include <unistd.h>
e39b673e 28#if !defined(__dietlibc__) && !defined(__KLIBC__)
64c4757e 29extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
f783ca4f 30#else
fffdbe5e 31# if defined(__NO_STAT64) || __WORDSIZE != 32
f783ca4f
NB
32# define lseek64 lseek
33# endif
98c6faba 34#endif
64c4757e
NB
35
36#include <sys/types.h>
37#include <sys/stat.h>
38#include <stdlib.h>
39#include <time.h>
11018a45 40#include <sys/time.h>
64c4757e
NB
41#include <getopt.h>
42#include <fcntl.h>
43#include <stdio.h>
44#include <errno.h>
45#include <string.h>
773135f5 46#include <syslog.h>
b56c3630
NB
47#ifdef __dietlibc__
48#include <strings.h>
280a927d
NB
49/* dietlibc has deprecated random and srandom!! */
50#define random rand
51#define srandom srand
98c6faba
NB
52#endif
53
64c4757e
NB
54
55#include <linux/kdev_t.h>
e0d19036
NB
56/*#include <linux/fs.h> */
57#include <sys/mount.h>
58#include <asm/types.h>
64c4757e
NB
59#include <sys/ioctl.h>
60#define MD_MAJOR 9
dd0781e5 61#define MdpMinorShift 6
64c4757e 62
e0d19036 63#ifndef BLKGETSIZE64
98c6faba 64#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 65#endif
56eb10c0 66
c21e737b 67#define DEFAULT_CHUNK 512
c82f047c
NB
68#define DEFAULT_BITMAP_CHUNK 4096
69#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 70#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e 71
753cf905 72/* MAP_DIR should be somewhere that persists across the pivotroot
5d4d1b26 73 * from early boot to late boot.
96fd06ed 74 * /run seems to have emerged as the best standard.
5d4d1b26 75 */
753cf905 76#ifndef MAP_DIR
96fd06ed 77#define MAP_DIR "/run/mdadm"
753cf905
DL
78#endif /* MAP_DIR */
79/* MAP_FILE is what we name the map file we put in MAP_DIR, in case you
80 * want something other than the default of "map"
81 */
82#ifndef MAP_FILE
83#define MAP_FILE "map"
84#endif /* MAP_FILE */
85/* MDMON_DIR is where pid and socket files used for communicating
96fd06ed 86 * with mdmon normally live. Best is /var/run/mdadm as
753cf905
DL
87 * mdmon is needed at early boot then it needs to write there prior
88 * to /var/run being mounted read/write, and it also then needs to
89 * persist beyond when /var/run is mounter read-only. So, to be
90 * safe, the default is somewhere that is read/write early in the
91 * boot process and stays up as long as possible during shutdown.
92 */
93#ifndef MDMON_DIR
96fd06ed 94#define MDMON_DIR "/run/mdadm"
753cf905 95#endif /* MDMON_DIR */
5d4d1b26 96
403410eb
PC
97/* FAILED_SLOTS is where to save files storing recent removal of array
98 * member in order to allow future reuse of disk inserted in the same
99 * slot for array recovery
100 */
101#ifndef FAILED_SLOTS_DIR
96fd06ed 102#define FAILED_SLOTS_DIR "/run/mdadm/failed-slots"
403410eb
PC
103#endif /* FAILED_SLOTS */
104
64c4757e 105#include "md_u.h"
e0d19036 106#include "md_p.h"
c82f047c 107#include "bitmap.h"
f7dd881f 108#include "msg.h"
64c4757e 109
1e0d770c 110#include <endian.h>
efd441d1
NB
111/* Redhat don't like to #include <asm/byteorder.h>, and
112 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
113 * and there is no standard conversion function so... */
0ae03b8a
NB
114/* And dietlibc doesn't think byteswap is ok, so.. */
115/* #include <byteswap.h> */
116#define bswap_16(x) (((x) & 0x00ffU) << 8 | \
117 ((x) & 0xff00U) >> 8)
118#define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
119 ((x) & 0xff000000U) >> 24 | \
120 ((x) & 0x0000ff00U) << 8 | \
121 ((x) & 0x00ff0000U) >> 8)
122#define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
123 ((x) & 0xff00000000000000ULL) >> 56 | \
124 ((x) & 0x000000000000ff00ULL) << 40 | \
125 ((x) & 0x00ff000000000000ULL) >> 40 | \
126 ((x) & 0x0000000000ff0000ULL) << 24 | \
127 ((x) & 0x0000ff0000000000ULL) >> 24 | \
128 ((x) & 0x00000000ff000000ULL) << 8 | \
129 ((x) & 0x000000ff00000000ULL) >> 8)
130
e39b673e 131#if !defined(__KLIBC__)
efd441d1
NB
132#if BYTE_ORDER == LITTLE_ENDIAN
133#define __cpu_to_le16(_x) (_x)
134#define __cpu_to_le32(_x) (_x)
135#define __cpu_to_le64(_x) (_x)
136#define __le16_to_cpu(_x) (_x)
137#define __le32_to_cpu(_x) (_x)
138#define __le64_to_cpu(_x) (_x)
974e620d
NB
139
140#define __cpu_to_be16(_x) bswap_16(_x)
141#define __cpu_to_be32(_x) bswap_32(_x)
142#define __cpu_to_be64(_x) bswap_64(_x)
143#define __be16_to_cpu(_x) bswap_16(_x)
144#define __be32_to_cpu(_x) bswap_32(_x)
145#define __be64_to_cpu(_x) bswap_64(_x)
efd441d1
NB
146#elif BYTE_ORDER == BIG_ENDIAN
147#define __cpu_to_le16(_x) bswap_16(_x)
148#define __cpu_to_le32(_x) bswap_32(_x)
149#define __cpu_to_le64(_x) bswap_64(_x)
150#define __le16_to_cpu(_x) bswap_16(_x)
151#define __le32_to_cpu(_x) bswap_32(_x)
152#define __le64_to_cpu(_x) bswap_64(_x)
974e620d
NB
153
154#define __cpu_to_be16(_x) (_x)
155#define __cpu_to_be32(_x) (_x)
156#define __cpu_to_be64(_x) (_x)
157#define __be16_to_cpu(_x) (_x)
158#define __be32_to_cpu(_x) (_x)
159#define __be64_to_cpu(_x) (_x)
1e0d770c
NB
160#else
161# error "unknown endianness."
162#endif
e39b673e 163#endif /* __KLIBC__ */
1e0d770c
NB
164
165
1e5c6983
DW
166/*
167 * min()/max()/clamp() macros that also do
168 * strict type-checking.. See the
169 * "unnecessary" pointer comparison.
170 */
171#define min(x, y) ({ \
172 typeof(x) _min1 = (x); \
173 typeof(y) _min2 = (y); \
174 (void) (&_min1 == &_min2); \
175 _min1 < _min2 ? _min1 : _min2; })
176
177#define max(x, y) ({ \
178 typeof(x) _max1 = (x); \
179 typeof(y) _max2 = (y); \
180 (void) (&_max1 == &_max2); \
181 _max1 > _max2 ? _max1 : _max2; })
1e0d770c 182
4b1ac34b
NB
183/* general information that might be extracted from a superblock */
184struct mdinfo {
185 mdu_array_info_t array;
186 mdu_disk_info_t disk;
187 __u64 events;
fbf8a0b7 188 int uuid[4];
31317663 189 char name[33];
353632d9 190 unsigned long long data_offset;
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) */
06c7f68e 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;
058574b1 260 struct supertype *supertype;
5bbb4842
NB
261};
262
9a9dab36 263#define Name "mdadm"
682c7051 264
e0d19036
NB
265enum mode {
266 ASSEMBLE=1,
267 BUILD,
268 CREATE,
269 MANAGE,
270 MISC,
271 MONITOR,
dd0781e5 272 GROW,
8382f19b 273 INCREMENTAL,
1f48664b 274 AUTODETECT,
5187a385 275 mode_count
e0d19036
NB
276};
277
64c4757e 278extern char short_options[];
024768c4 279extern char short_bitmap_options[];
c06487ce 280extern char short_bitmap_auto_options[];
64c4757e 281extern struct option long_options[];
56eedc1a 282extern char Version[], Usage[], Help[], OptionHelp[],
5187a385 283 *mode_help[],
dd0781e5 284 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 285 Help_incr[],
e0d19036 286 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 287
997aed5d 288/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 289 * numbers later than any 'short' character option.
997aed5d
NB
290 */
291enum special_options {
1c7a808c 292 AssumeClean = 300,
997aed5d
NB
293 BitmapChunk,
294 WriteBehind,
295 ReAdd,
296 NoDegraded,
297 Sparc22,
1c7a808c 298 BackupFile,
997aed5d 299 HomeHost,
589395d6 300 AutoHomeHost,
38098016 301 Symlinks,
1f48664b 302 AutoDetect,
1770662b 303 Waitclean,
4cce4069 304 DetailPlatform,
33414a01 305 KillSubarray,
1c7a808c 306 UpdateSubarray,
edde9560 307 IncrementalPath,
1c7a808c
N
308 NoSharing,
309 HelpOptions,
310 Brief,
311 ManageOpt,
312 Add,
313 Remove,
314 Fail,
315 MiscOpt,
316 WaitOpt,
317 ConfigFile,
318 ChunkSize,
319 WriteMostly,
320 Layout,
321 Auto,
322 Force,
323 SuperMinor,
324 EMail,
325 ProgramOpt,
326 Increment,
327 Fork,
328 Bitmap,
329 RebuildMapOpt,
87f26d14 330 InvalidBackup,
20b60dcd 331 UdevRules,
b76b30e0 332 FreezeReshape,
2dddadb0 333 Continue,
08ca2adf 334 OffRootOpt,
c2ecf5f6 335 Prefer,
f7d3febc 336 KillOpt,
40c9a66a 337 DataOffset,
997aed5d
NB
338};
339
f0ec6710
MN
340enum prefix_standard {
341 JEDEC,
342 IEC
343};
344
64c4757e 345/* structures read from config file */
52826846
NB
346/* List of mddevice names and identifiers
347 * Identifiers can be:
348 * uuid=128-hex-uuid
349 * super-minor=decimal-minor-number-from-superblock
350 * devices=comma,separated,list,of,device,names,with,wildcards
351 *
352 * If multiple fields are present, the intersection of all matching
353 * devices is considered
354 */
98c6faba 355#define UnSet (0xfffe)
fa56eddb 356struct mddev_ident {
dd0781e5 357 char *devname;
aba69144 358
dd0781e5 359 int uuid_set;
3fa06e9d 360 int uuid[4];
947fd4dd 361 char name[33];
52826846 362
f21e18ca 363 int super_minor;
52826846 364
dd0781e5 365 char *devices; /* comma separated list of device
52826846
NB
366 * names with wild cards
367 */
dd0781e5 368 int level;
f21e18ca
N
369 int raid_disks;
370 int spare_disks;
82d9eba6 371 struct supertype *st;
dd0781e5
NB
372 int autof; /* 1 for normal, 2 for partitioned */
373 char *spare_group;
7ef02d01 374 char *bitmap_file;
c82f047c 375 int bitmap_fd;
dd0781e5 376
1771a6e2
N
377 char *container; /* /dev/whatever name of container, or
378 * uuid of container. You would expect
379 * this to be the 'devname' or UUID
dbb44303
N
380 * of some other entry.
381 */
382 char *member; /* subarray within a container */
383
fa56eddb 384 struct mddev_ident *next;
b179246f
N
385 union {
386 /* fields needed by different users of this structure */
387 int assembled; /* set when assembly succeeds */
388 };
fa56eddb 389};
64c4757e 390
9e33d556
N
391struct context {
392 int readonly;
393 int runstop;
394 int verbose;
9e33d556
N
395 int brief;
396 int force;
397 char *homehost;
398 int require_homehost;
399 char *prefer;
400 int export;
401 int test;
402 char *subarray;
403 char *update;
404 int scan;
405 int SparcAdjust;
406 int autof;
407 int delay;
408 int freeze_reshape;
409 char *backup_file;
410 int invalid_backup;
411};
412
e705e81b
N
413struct shape {
414 int raiddisks;
415 int sparedisks;
416 int level;
417 int layout;
418 char *layout_str;
419 int chunk;
420 int bitmap_chunk;
421 char *bitmap_file;
422 int assume_clean;
423 int write_behind;
424 unsigned long long size;
425};
426
64c4757e 427/* List of device names - wildcards expanded */
a655e550 428struct mddev_dev {
64c4757e 429 char *devname;
c8e1a230
N
430 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail,
431 * 'A' for re_add.
cd29a5c8
NB
432 * Not set for names read from .config
433 */
b3d31955 434 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
811e6cbe 435 char used; /* set when used */
a655e550
N
436 struct mddev_dev *next;
437};
64c4757e 438
682c7051
NB
439typedef struct mapping {
440 char *name;
441 int num;
442} mapping_t;
443
e0d19036
NB
444
445struct mdstat_ent {
446 char *dev;
447 int devnum;
448 int active;
449 char *level;
450 char *pattern; /* U or up, _ for down */
451 int percent; /* -1 if no resync */
f94c116f 452 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
453 int devcnt;
454 int raid_disks;
549e9569 455 char * metadata_version;
3b57c466
N
456 struct dev_member {
457 char *name;
458 struct dev_member *next;
459 } *members;
e0d19036
NB
460 struct mdstat_ent *next;
461};
462
22a88995 463extern struct mdstat_ent *mdstat_read(int hold, int start);
e0d19036 464extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 465extern void mdstat_wait(int seconds);
58a4ba2a 466extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
8382f19b 467extern int mddev_busy(int devnum);
3b57c466 468extern struct mdstat_ent *mdstat_by_component(char *name);
78b10e66 469extern struct mdstat_ent *mdstat_by_subdev(char *subdev, int container);
8382f19b
NB
470
471struct map_ent {
472 struct map_ent *next;
473 int devnum;
1522c538 474 char metadata[20];
8382f19b 475 int uuid[4];
195254b8 476 int bad;
8382f19b
NB
477 char *path;
478};
1522c538 479extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
8382f19b 480 int uuid[4], char *path);
4eb26970 481extern void map_remove(struct map_ent **map, int devnum);
8382f19b 482extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
f2e55ecc
N
483extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
484extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
485extern void map_read(struct map_ent **melp);
486extern int map_write(struct map_ent *mel);
487extern void map_delete(struct map_ent **mapp, int devnum);
488extern void map_free(struct map_ent *map);
489extern void map_add(struct map_ent **melp,
1522c538 490 int devnum, char *metadata, int uuid[4], char *path);
ad5bc697
N
491extern int map_lock(struct map_ent **melp);
492extern void map_unlock(struct map_ent **melp);
cc700db3 493extern void map_fork(void);
e0d19036 494
e86c9dd6 495/* various details can be requested */
dab4a513
DW
496enum sysfs_read_flags {
497 GET_LEVEL = (1 << 0),
498 GET_LAYOUT = (1 << 1),
499 GET_COMPONENT = (1 << 2),
500 GET_CHUNK = (1 << 3),
501 GET_CACHE = (1 << 4),
502 GET_MISMATCH = (1 << 5),
503 GET_VERSION = (1 << 6),
504 GET_DISKS = (1 << 7),
505 GET_DEGRADED = (1 << 8),
506 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
507 GET_BITMAP_LOCATION = (1 << 10),
508
509 GET_DEVS = (1 << 20), /* gets role, major, minor */
510 GET_OFFSET = (1 << 21),
511 GET_SIZE = (1 << 22),
512 GET_STATE = (1 << 23),
513 GET_ERROR = (1 << 24),
dab4a513 514};
e86c9dd6
NB
515
516/* If fd >= 0, get the array it is open on,
517 * else use devnum. >=0 -> major9. <0.....
518 */
549e9569 519extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 520extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
521extern void sysfs_free(struct mdinfo *sra);
522extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
523extern int sysfs_attr_match(const char *attr, const char *str);
524extern int sysfs_match_word(const char *word, char **list);
7e0f6979 525extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 526 char *name, char *val);
7e0f6979 527extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 528 char *name, unsigned long long val);
012a8641
JS
529extern int sysfs_set_num_signed(struct mdinfo *sra, struct mdinfo *dev,
530 char *name, long long val);
97590376 531extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
532extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
533 char *name);
534extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 535extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 536 char *name, unsigned long long *val);
7236ee7a 537extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
538extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
539 char *name);
93ecfa01
N
540extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
541 char *name, char *val, int size);
8ed3e5e1 542extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 543extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 544extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 545extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 546extern int sysfs_unique_holder(int devnum, long rdev);
bc77ed53 547extern int sysfs_freeze_array(struct mdinfo *sra);
c69b251b 548extern int load_sys(char *path, char *buf);
130994cb
AK
549extern int reshape_prepare_fdlist(char *devname,
550 struct mdinfo *sra,
551 int raid_disks,
552 int nrdisks,
553 unsigned long blocks,
554 char *backup_file,
555 int *fdlist,
556 unsigned long long *offsets);
557extern void reshape_free_fdlist(int *fdlist,
558 unsigned long long *offsets,
559 int size);
e6e9d47b
AK
560extern int reshape_open_backup_file(char *backup,
561 int fd,
562 char *devname,
563 long blocks,
564 int *fdlist,
a93f87ee
N
565 unsigned long long *offsets,
566 int restart);
1c009fc2
AK
567extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
568 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
569
570extern int save_stripes(int *source, unsigned long long *offsets,
571 int raid_disks, int chunk_size, int level, int layout,
572 int nwrites, int *dest,
a6288483
N
573 unsigned long long start, unsigned long long length,
574 char *buf);
353632d9
NB
575extern int restore_stripes(int *dest, unsigned long long *offsets,
576 int raid_disks, int chunk_size, int level, int layout,
577 int source, unsigned long long read_offset,
2fcb75ae
AK
578 unsigned long long start, unsigned long long length,
579 char *src_buf);
e86c9dd6 580
52826846
NB
581#ifndef Sendmail
582#define Sendmail "/usr/lib/sendmail -t"
583#endif
584
773135f5
NB
585#define SYSLOG_FACILITY LOG_DAEMON
586
682c7051
NB
587extern char *map_num(mapping_t *map, int num);
588extern int map_name(mapping_t *map, char *name);
b640a252 589extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 590
c2ecf5f6
N
591extern char *map_dev_preferred(int major, int minor, int create,
592 char *prefer);
593static inline char *map_dev(int major, int minor, int create)
594{
595 return map_dev_preferred(major, minor, create, NULL);
596}
64c4757e 597
549e9569 598struct active_array;
2e735d19 599struct metadata_update;
64c4757e 600
999b4972 601
80bf9135 602/* 'struct reshape' records the intermediate states of
999b4972
N
603 * a general reshape.
604 * The starting geometry is converted to the 'before' geometry
605 * by at most an atomic level change. They could be the same.
606 * Similarly the 'after' geometry is converted to the final
607 * geometry by at most a level change.
608 * Note that 'before' and 'after' must have the same level.
609 * 'blocks' is the minimum number of sectors for a reshape unit.
610 * This will be a multiple of the stripe size in each of the
611 * 'before' and 'after' geometries.
612 * If 'blocks' is 0, no restriping is necessary.
613 */
614struct reshape {
615 int level;
616 int parity; /* number of parity blocks/devices */
617 struct {
618 int layout;
619 int data_disks;
620 } before, after;
621 unsigned long long backup_blocks;
622 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
623 unsigned long long new_size; /* New size of array in sectors */
624};
625
6adfd3af
NB
626/* A superswitch provides entry point the a metadata handler.
627 *
628 * The super_switch primarily operates on some "metadata" that
629 * is accessed via the 'supertype'.
630 * This metadata has one of three possible sources.
631 * 1/ It is read from a single device. In this case it may not completely
632 * describe the array or arrays as some information might be on other
633 * devices.
634 * 2/ It is read from all devices in a container. In this case all
635 * information is present.
636 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
637 * be complete once enough ->add_to_super calls have completed.
638 *
639 * When creating an array inside a container, the metadata will be
640 * formed by a combination of 2 and 3. The metadata or the array is read,
641 * then new information is added.
642 *
643 * The metadata must sometimes have a concept of a 'current' array
644 * and a 'current' device.
645 * The 'current' array is set by init_super to be the newly created array,
646 * or is set by super_by_fd when it finds it is looking at an array inside
647 * a container.
648 *
649 * The 'current' device is either the device that the metadata was read from
650 * in case 1, or the last device added by add_to_super in case 3.
651 * Case 2 does not identify a 'current' device.
652 */
f9ce90ba 653extern struct superswitch {
6adfd3af
NB
654
655 /* Used to report details of metadata read from a component
656 * device. ->load_super has been called.
657 */
3da92f27 658 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 659 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 660 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 661 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
662
663 /* Used to report details of an active array.
664 * ->load_super was possibly given a 'component' string.
665 */
3da92f27 666 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 667 void (*brief_detail_super)(struct supertype *st);
0d726f17 668 void (*export_detail_super)(struct supertype *st);
6adfd3af 669
4cce4069 670 /* Optional: platform hardware / firmware details */
9eafa1de
MN
671 int (*detail_platform)(int verbose, int enumerate_only, char *controller_path);
672 int (*export_detail_platform)(int verbose, char *controller_path);
4cce4069 673
6adfd3af
NB
674 /* Used:
675 * to get uuid to storing in bitmap metadata
676 * and 'reshape' backup-data metadata
677 * To see if a device is being re-added to an array it was part of.
678 */
3da92f27 679 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 680
8592f29d 681 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
682 * the container, or about an individual array within the container.
683 * The determination is made either by:
684 * load_super being given a 'component' string.
685 * validate_geometry determining what to create.
d2ca6449
NB
686 * The info includes both array information and device information.
687 * The particular device should be:
688 * The last device added by add_to_super
689 * The device the metadata was loaded from by load_super
a5d85af7
N
690 * If 'map' is present, then it is an array raid_disks long
691 * (raid_disk must already be set and correct) and it is filled
692 * with 1 for slots that are thought to be active and 0 for slots which
693 * appear to be failed/missing.
95eeceeb 694 * *info is zeroed out before data is added.
6adfd3af 695 */
a5d85af7 696 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 697 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 698 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 699 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 700 */
3da92f27 701 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
702
703 /* Make one of several generic modifications to metadata
704 * prior to assembly (or other times).
705 * sparc2.2 - first bug in early 0.90 metadata
706 * super-minor - change name of 0.90 metadata
707 * summaries - 'correct' any redundant data
708 * resync - mark array as dirty to trigger a resync.
709 * uuid - set new uuid - only 0.90 or 1.x
710 * name - change the name of the array (where supported)
711 * homehost - change which host this array is tied to.
712 * devicesize - If metadata is at start of device, change recorded
713 * device size to match actual device size
714 * byteorder - swap bytes for 0.90 metadata
715 *
716 * force-one - mark that device as uptodate, not old or failed.
717 * force-array - mark array as clean if it would not otherwise
718 * assemble
719 * assemble - not sure how this is different from force-one...
720 * linear-grow-new - add a new device to a linear array, but don't
721 * change the size: so superblock still matches
722 * linear-grow-update - now change the size of the array.
16715c01
DL
723 * writemostly - set the WriteMostly1 bit in the superblock devflags
724 * readwrite - clear the WriteMostly1 bit in the superblock devflags
6adfd3af 725 */
68c7d6d7 726 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 727 char *update,
e5eac01f
NB
728 char *devname, int verbose,
729 int uuid_set, char *homehost);
6adfd3af
NB
730
731 /* Create new metadata for new array as described. This could
732 * be a new container, or an array in a pre-existing container.
733 * Also used to zero metadata prior to writing it to invalidate old
734 * metadata.
735 */
3da92f27
NB
736 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
737 unsigned long long size, char *name,
83cd1e97
N
738 char *homehost, int *uuid,
739 unsigned long long data_offset);
6adfd3af
NB
740
741 /* update the metadata to include new device, either at create or
742 * when hot-adding a spare.
743 */
f20c3968 744 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
1a64be56
LM
745 int fd, char *devname);
746 /* update the metadata to delete a device,
747 * when hot-removing.
748 */
749 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
750
751 /* Write metadata to one device when fixing problems or adding
752 * a new device.
753 */
3da92f27 754 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
755
756 /* Write all metadata for this array.
757 */
111d01fc 758 int (*write_init_super)(struct supertype *st);
64557c33 759 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 760 int (*load_super)(struct supertype *st, int fd, char *devname);
2b959fbf 761 int (*load_container)(struct supertype *st, int fd, char *devname);
82d9eba6 762 struct supertype * (*match_metadata_desc)(char *arg);
387fcd59
N
763 __u64 (*avail_size)(struct supertype *st, __u64 size,
764 unsigned long long data_offset);
80e7f8c3 765 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
3da92f27 766 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 767 int delay, int write_behind,
f9c25f1d 768 unsigned long long size, int may_change, int major);
3da92f27
NB
769 void (*locate_bitmap)(struct supertype *st, int fd);
770 int (*write_bitmap)(struct supertype *st, int fd);
771 void (*free_super)(struct supertype *st);
78e44928
NB
772
773 /* validate_geometry is called with an st returned by
774 * match_metadata_desc.
775 * It should check that the geometry described in compatible with
776 * the metadata type. It will be called repeatedly as devices
777 * added to validate changing size and new devices. If there are
778 * inter-device dependencies, it should record sufficient details
779 * so these can be validated.
30f58b22 780 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
781 * Return value is:
782 * 1: everything is OK
783 * 0: not OK for some reason - if 'verbose', then error was reported.
784 * -1: st->sb was NULL, 'subdev' is a member of a container of this
785 * types, but array is not acceptable for some reason
786 * message was reported even if verbose is 0.
78e44928 787 */
17f25ca6
NB
788 int (*validate_geometry)(struct supertype *st, int level, int layout,
789 int raiddisks,
c21e737b 790 int *chunk, unsigned long long size,
af4348dd 791 unsigned long long data_offset,
2c514b71
NB
792 char *subdev, unsigned long long *freesize,
793 int verbose);
598f0d58 794
00bbdbda 795 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
796 /* query the supertype for default geometry */
797 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
798 /* Permit subarray's to be deleted from inactive containers */
799 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 800 /* Permit subarray's to be modified */
a951a4f7 801 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 802 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
803 /* Check if reshape is supported for this external format.
804 * st is obtained from super_by_fd() where st->subarray[0] is
805 * initialized to indicate if reshape is being performed at the
806 * container or subarray level
807 */
016e00f5
AK
808#define APPLY_METADATA_CHANGES 1
809#define ROLLBACK_METADATA_CHANGES 0
810
d04f65f4
N
811 int (*reshape_super)(struct supertype *st,
812 unsigned long long size, int level,
7bc71196 813 int layout, int chunksize, int raid_disks,
41784c88 814 int delta_disks, char *backup, char *dev,
016e00f5 815 int direction,
41784c88 816 int verbose); /* optional */
999b4972
N
817 int (*manage_reshape)( /* optional */
818 int afd, struct mdinfo *sra, struct reshape *reshape,
819 struct supertype *st, unsigned long blocks,
820 int *fds, unsigned long long *offsets,
821 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 822
549e9569 823/* for mdmon */
cba0191b
NB
824 int (*open_new)(struct supertype *c, struct active_array *a,
825 char *inst);
ed9d66aa
NB
826
827 /* Tell the metadata handler the current state of the array.
828 * This covers whether it is known to be consistent (no pending writes)
01f157d7 829 * and how far along a resync is known to have progressed
ed9d66aa
NB
830 * (in a->resync_start).
831 * resync status is really irrelevant if the array is not consistent,
832 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
833 * If 'consistent' is '2', then the array can mark it dirty if a
834 * resync/recovery/whatever is required, or leave it clean if not.
835 * Return value is 0 dirty (not consistent) and 1 if clean.
836 * it is only really important if consistent is passed in as '2'.
4e5528c6 837 */
01f157d7 838 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
839
840 /* When the state of a device might have changed, we call set_disk to
841 * tell the metadata what the current state is.
842 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
843 * transitions.
844 * set_disk might be called when the state of the particular disk has
845 * not in fact changed.
846 */
8d45d196 847 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
848 void (*sync_metadata)(struct supertype *st);
849 void (*process_update)(struct supertype *st,
850 struct metadata_update *update);
edd8d13c
NB
851 void (*prepare_update)(struct supertype *st,
852 struct metadata_update *update);
549e9569 853
6c3fb95c
NB
854 /* activate_spare will check if the array is degraded and, if it
855 * is, try to find some spare space in the container.
856 * On success, it add appropriate updates (For process_update) to
857 * to the 'updates' list and returns a list of 'mdinfo' identifying
858 * the device, or devices as there might be multiple missing
859 * devices and multiple spares available.
860 */
861 struct mdinfo *(*activate_spare)(struct active_array *a,
862 struct metadata_update **updates);
2cda7640
ML
863 /*
864 * Return statically allocated string that represents metadata specific
865 * controller domain of the disk. The domain is used in disk domain
866 * matching functions. Disks belong to the same domain if the they have
867 * the same domain from mdadm.conf and belong the same metadata domain.
868 * Returning NULL or not providing this handler means that metadata
869 * does not distinguish the differences between disks that belong to
870 * different controllers. They are in the domain specified by
871 * configuration file (mdadm.conf).
872 * In case when the metadata has the notion of domains based on disk
873 * it shall return NULL for disks that do not belong to the controller
874 * the supported domains. Such disks will form another domain and won't
875 * be mixed with supported ones.
876 */
877 const char *(*get_disk_controller_domain)(const char *path);
549e9569 878
ba53ea59
AK
879 /* for external backup area */
880 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
881
f277ce36 882 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 883 int external;
4cce4069 884 const char *name; /* canonical metadata name */
0f22b998 885} *superlist[];
82d9eba6 886
0f22b998
N
887extern struct superswitch super0, super1;
888extern struct superswitch super_imsm, super_ddf;
0592faeb 889extern struct superswitch mbr, gpt;
cdddbdbc 890
edd8d13c
NB
891struct metadata_update {
892 int len;
893 char *buf;
894 void *space; /* allocated space that monitor will use */
cb23f1f4
N
895 void **space_list; /* list of allocated spaces that monitor can
896 * use or that it returned.
897 */
edd8d13c
NB
898 struct metadata_update *next;
899};
900
6adfd3af
NB
901/* A supertype holds a particular collection of metadata.
902 * It identifies the metadata type by the superswitch, and the particular
903 * sub-version of that metadata type.
904 * metadata read in or created is stored in 'sb' and 'info'.
905 * There are also fields used by mdmon to track containers.
906 *
352452c3
N
907 * A supertype may refer to:
908 * Just an array, possibly in a container
909 * A container, not identifying any particular array
910 * Info read from just one device, not yet fully describing the array/container.
911 *
912 *
6adfd3af
NB
913 * A supertype is created by:
914 * super_by_fd
915 * guess_super
916 * dup_super
917 */
82d9eba6
NB
918struct supertype {
919 struct superswitch *ss;
920 int minor_version;
ea329559 921 int max_devs;
2f6079dc 922 int container_dev; /* devnum of container */
64557c33 923 void *sb;
111d01fc 924 void *info;
df3346e6
LM
925 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
926 HW/firmware related incompatability to load metadata.
927 Used when examining metadata to display content of disk
928 when user has no hw/firmare compatible system.
929 */
edd8d13c
NB
930 struct metadata_update *updates;
931 struct metadata_update **update_tail;
932
549e9569
NB
933 /* extra stuff used by mdmon */
934 struct active_array *arrays;
549e9569 935 int sock; /* listen to external programs */
549e9569
NB
936 int devnum;
937 char *devname; /* e.g. md0. This appears in metadata_verison:
938 * external:/md0/12
939 */
940 int devcnt;
68226a80 941 int retry_soon;
549e9569
NB
942
943 struct mdinfo *devs;
944
82d9eba6 945};
f9ce90ba 946
4725bc31 947extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
948enum guess_types { guess_any, guess_array, guess_partitions };
949extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
950static inline struct supertype *guess_super(int fd) {
951 return guess_super_type(fd, guess_any);
952}
3da92f27 953extern struct supertype *dup_super(struct supertype *st);
beae1dfe 954extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 955extern int must_be_container(int fd);
0f0749ad 956extern int dev_size_from_id(dev_t id, unsigned long long *size);
8382f19b
NB
957extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
958 mdu_disk_info_t *disk);
a7c6e3fb 959void wait_for(char *dev, int fd);
f9ce90ba 960
5527fc74
N
961/*
962 * Data structures for policy management.
963 * Each device can have a policy structure that lists
964 * various name/value pairs each possibly with a metadata associated.
965 * The policy list is sorted by name/value/metadata
966 */
967struct dev_policy {
968 struct dev_policy *next;
969 char *name; /* None of these strings are allocated. They are
970 * all just references to strings which are known
971 * to exist elsewhere.
972 * name and metadata can be compared by address equality.
973 */
974 const char *metadata;
2cda7640 975 const char *value;
5527fc74
N
976};
977
4e8d9f0a 978extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
979
980/* iterate over the sublist starting at list, having the same
981 * 'name' as 'list', and matching the given metadata (Where
982 * NULL matches anything
983 */
984#define pol_for_each(item, list, _metadata) \
985 for (item = list; \
986 item && item->name == list->name; \
987 item = item->next) \
988 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
989 ; else
990
991/*
992 * policy records read from mdadm are largely just name-value pairs.
993 * The names are constants, not strdupped
994 */
995struct pol_rule {
996 struct pol_rule *next;
997 char *type; /* rule_policy or rule_part */
998 struct rule {
999 struct rule *next;
1000 char *name;
1001 char *value;
1002 char *dups; /* duplicates of 'value' with a partNN appended */
1003 } *rule;
1004};
1005
1006extern char rule_policy[], rule_part[];
1007extern char rule_path[], rule_type[];
1008extern char type_part[], type_disk[];
1009
1010extern void policyline(char *line, char *type);
4e8d9f0a 1011extern void policy_add(char *type, ...);
5527fc74
N
1012extern void policy_free(void);
1013
73c9c47c 1014extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 1015extern struct dev_policy *disk_policy(struct mdinfo *disk);
4e8d9f0a 1016extern struct dev_policy *devnum_policy(int dev);
5527fc74
N
1017extern void dev_policy_free(struct dev_policy *p);
1018
e78dda3b
N
1019//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
1020extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
1021extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
1022
e3bb5f14
N
1023enum policy_action {
1024 act_default,
1025 act_include,
1026 act_re_add,
d2db3045
N
1027 act_spare, /* This only applies to bare devices */
1028 act_spare_same_slot, /* this allows non-bare devices,
1029 * but only if recent removal */
1030 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
1031 act_err
1032};
1033
1034extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
1035 enum policy_action want);
1036extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
1037 enum policy_action want);
f5f12c84
N
1038
1039struct domainlist {
1040 struct domainlist *next;
2cda7640 1041 const char *dom;
f5f12c84
N
1042};
1043
1044extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
1045 const char *metadata);
1046extern struct domainlist *domain_from_array(struct mdinfo *mdi,
1047 const char *metadata);
e78dda3b
N
1048extern void domainlist_add_dev(struct domainlist **dom, int devnum,
1049 const char *metadata);
f5f12c84
N
1050extern void domain_free(struct domainlist *dl);
1051extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
1052 const char *metadata);
e78dda3b 1053void domain_add(struct domainlist **domp, char *domain);
f5f12c84 1054
403410eb
PC
1055extern void policy_save_path(char *id_path, struct map_ent *array);
1056extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
1057
838acbc2
NB
1058#if __GNUC__ < 3
1059struct stat64;
1060#endif
1061
45e878bb
NB
1062#define HAVE_NFTW we assume
1063#define HAVE_FTW
1064
2df1f269 1065#ifdef __UCLIBC__
45e878bb 1066# include <features.h>
2df1f269
BRF
1067# ifndef __UCLIBC_HAS_LFS__
1068# define lseek64 lseek
1069# endif
45e878bb
NB
1070# ifndef __UCLIBC_HAS_FTW__
1071# undef HAVE_FTW
1072# undef HAVE_NFTW
1073# endif
1074#endif
1075
1076#ifdef __dietlibc__
1077# undef HAVE_NFTW
1078#endif
1079
e39b673e 1080#if defined(__KLIBC__)
1081# undef HAVE_NFTW
1082# undef HAVE_FTW
1083#endif
1084
45e878bb 1085#ifndef HAVE_NFTW
173fc515 1086# define FTW_PHYS 1
45e878bb
NB
1087# ifndef HAVE_FTW
1088 struct FTW {};
173fc515
NB
1089# endif
1090#endif
1091
45e878bb
NB
1092#ifdef HAVE_FTW
1093# include <ftw.h>
1094#endif
1095
173fc515
NB
1096extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1097
f9ce90ba 1098
64c4757e 1099extern int Manage_ro(char *devname, int fd, int readonly);
ba728be7
N
1100extern int Manage_runstop(char *devname, int fd, int runstop, int quiet,
1101 int will_retry);
64c4757e 1102extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1103 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1104 char *update, int force);
1f48664b 1105extern int autodetect(void);
e5329c37 1106extern int Grow_Add_device(char *devname, int fd, char *newdev);
50f01ba5
N
1107extern int Grow_addbitmap(char *devname, int fd,
1108 struct context *c, struct shape *s);
32754b7d 1109extern int Grow_reshape(char *devname, int fd,
e2e53a2d 1110 struct mddev_dev *devlist,
40c9a66a 1111 unsigned long long data_offset,
32754b7d 1112 struct context *c, struct shape *s);
06b0d786 1113extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1114 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1115extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1116 struct mdinfo *info, char *backup_file,
1117 int freeze_reshape);
1118
3f54bd62
AK
1119extern int restore_backup(struct supertype *st,
1120 struct mdinfo *content,
1121 int working_disks,
1122 int spares,
1123 char *backup_file,
1124 int verbose);
2dddadb0
AK
1125extern int Grow_continue_command(char *devname, int fd,
1126 char *backup_file, int verbose);
64c4757e 1127
7f91af49 1128extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1129 struct mddev_ident *ident,
87f26d14 1130 struct mddev_dev *devlist,
4977146a 1131 struct context *c);
64c4757e 1132
a4e8316a
N
1133extern int Build(char *mddev, struct mddev_dev *devlist,
1134 struct shape *s, struct context *c);
64c4757e 1135
7f91af49 1136extern int Create(struct supertype *st, char *mddev,
171dccc8 1137 char *name, int *uuid,
a655e550 1138 int subdevs, struct mddev_dev *devlist,
99cc42f4 1139 struct shape *s,
40c9a66a
N
1140 struct context *c,
1141 unsigned long long data_offset);
64c4757e 1142
ef898ce6 1143extern int Detail(char *dev, struct context *c);
9eafa1de 1144extern int Detail_Platform(struct superswitch *ss, int scan, int verbose, int export, char *controller_path);
e0d19036 1145extern int Query(char *dev);
eec3f887
N
1146extern int Examine(struct mddev_dev *devlist, struct context *c,
1147 struct supertype *forcest);
a655e550 1148extern int Monitor(struct mddev_dev *devlist,
e0d19036 1149 char *mailaddr, char *alert_cmd,
95c50205
N
1150 struct context *c,
1151 int daemonise, int oneshot,
1152 int dosyslog, char *pidfile, int increments,
1153 int share);
64c4757e 1154
ba728be7
N
1155extern int Kill(char *dev, struct supertype *st, int force, int verbose, int noexcl);
1156extern int Kill_subarray(char *dev, char *subarray, int verbose);
fa56eddb 1157extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1158extern int Wait(char *dev);
9f1da824 1159extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1160
11b6d91d
N
1161extern int Incremental(char *devname, struct context *c,
1162 struct supertype *st);
8382f19b
NB
1163extern void RebuildMap(void);
1164extern int IncrementalScan(int verbose);
950bc344 1165extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1166extern int CreateBitmap(char *filename, int force, char uuid[16],
1167 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1168 unsigned long write_behind,
dcec9ee5
NB
1169 unsigned long long array_size,
1170 int major);
55935d51 1171extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1172extern int Write_rules(char *rule_name);
bf4fb153 1173extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1174extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 1175
64c4757e 1176extern int md_get_version(int fd);
e0d19036 1177extern int get_linux_version(void);
bc77ed53 1178extern int mdadm_version(char *version);
51d4261c 1179extern unsigned long long parse_size(char *size);
64c4757e 1180extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1181extern int parse_layout_10(char *layout);
1182extern int parse_layout_faulty(char *layout);
79868890 1183extern long parse_num(char *num);
682c7051
NB
1184extern int check_ext2(int fd, char *name);
1185extern int check_reiser(int fd, char *name);
1186extern int check_raid(int fd, char *name);
53ed6ac3
KW
1187extern int check_partitions(int fd, char *dname,
1188 unsigned long long freesize,
1189 unsigned long long size);
64c4757e 1190
dd0781e5 1191extern int get_mdp_major(void);
8b0dabea 1192extern int dev_open(char *dev, int flags);
e8a70c89 1193extern int open_dev(int devnum);
d998b738 1194extern int open_dev_flags(int devnum, int flags);
a931db9e 1195extern int open_dev_excl(int devnum);
8d80900b 1196extern int is_standard(char *dev, int *nump);
9008ed1c 1197extern int same_dev(char *one, char *two);
9eafa1de 1198extern int compare_paths (char* path1,char* path2);
dd0781e5 1199
f1ae21c4 1200extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1201extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1202extern struct mddev_dev *conf_get_devs(void);
8382f19b 1203extern int conf_test_dev(char *devname);
4e8d9f0a 1204extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1205extern struct createinfo *conf_get_create_info(void);
1206extern void set_conffile(char *file);
1207extern char *conf_get_mailaddr(void);
1208extern char *conf_get_mailfrom(void);
1209extern char *conf_get_program(void);
0ac91628 1210extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1211extern char *conf_line(FILE *file);
98c6faba 1212extern char *conf_word(FILE *file, int allow_key);
7103b9b8
N
1213extern void print_quoted(char *str);
1214extern void print_escape(char *str);
0ac91628 1215extern int conf_name_is_free(char *name);
7c336758 1216extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1217extern int devname_matches(char *name, char *match);
2244d1a9
N
1218extern struct mddev_ident *conf_match(struct supertype *st,
1219 struct mdinfo *info,
1220 char *devname,
1221 int verbose, int *rvp);
6d11ec6f 1222extern int experimental(void);
0ac91628 1223
e0d19036
NB
1224extern void free_line(char *line);
1225extern int match_oneof(char *devices, char *devname);
e0d19036 1226extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1227extern const int uuid_zero[4];
f277ce36 1228extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1229extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1230extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1231extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1232 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1233extern unsigned long calc_csum(void *super, int bytes);
583315d9 1234extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1235 char *avail);
8453e704 1236extern int enough_fd(int fd);
e0d19036 1237extern int ask(char *mesg);
8fac0577 1238extern unsigned long long get_component_size(int fd);
0430ed48 1239extern void remove_partitions(int fd);
691c6ee1 1240extern int test_partition(int fd);
bfd76b93 1241extern int test_partition_from_id(dev_t id);
577e8448 1242extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1243extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1244 int chunksize, unsigned long long devsize);
edd8d13c
NB
1245extern int flush_metadata_updates(struct supertype *st);
1246extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c 1247extern int assemble_container_content(struct supertype *st, int mdfd,
11b6d91d
N
1248 struct mdinfo *content,
1249 struct context *c,
1250 char *chosen_name);
326727d9
AC
1251extern struct mdinfo *container_choose_spares(struct supertype *st,
1252 unsigned long long min_size,
1253 struct domainlist *domlist,
1254 char *spare_group,
1255 const char *metadata, int get_one);
d52bb542 1256extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1257extern int add_disk(int mdfd, struct supertype *st,
1258 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1259extern int remove_disk(int mdfd, struct supertype *st,
1260 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1261extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1262unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1263
56eb10c0 1264extern char *human_size(long long bytes);
f0ec6710 1265extern char *human_size_brief(long long bytes, int prefix);
e4965ef8 1266extern void print_r10_layout(int layout);
e0d19036 1267
63152c1b
NB
1268#define NoMdDev (1<<23)
1269extern int find_free_devnum(int use_partitions);
1270
e0d19036
NB
1271extern void put_md_name(char *name);
1272extern char *get_md_name(int dev);
1273
c913b90e 1274extern char DefaultConfFile[];
b5e64645 1275
69207ff6
N
1276extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1277 char *chosen);
1278/* values for 'trustworthy' */
1279#define LOCAL 1
d1d3482b 1280#define LOCAL_ANY 10
69207ff6
N
1281#define FOREIGN 2
1282#define METADATA 3
6be1d39d 1283extern int open_mddev(char *dev, int report_errors);
a322f70c 1284extern int open_container(int fd);
33414a01
DW
1285extern int is_container_member(struct mdstat_ent *ent, char *devname);
1286extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1287extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1288extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1289
a931db9e 1290extern int mdmon_running(int devnum);
24f6f99b 1291extern int mdmon_pid(int devnum);
40ebbb9c 1292extern int check_env(char *name);
148acb7b 1293extern __u32 random32(void);
8850ee3e 1294extern int start_mdmon(int devnum);
a931db9e 1295
999b4972
N
1296extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1297 struct supertype *st, unsigned long stripes,
1298 int *fds, unsigned long long *offsets,
1299 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1300void abort_reshape(struct mdinfo *sra);
999b4972 1301
2f6079dc 1302extern char *devnum2devname(int num);
11877f4d 1303extern void fmt_devname(char *name, int num);
77472ff8 1304extern int devname2devnum(char *name);
c94709e8 1305extern int stat2devnum(struct stat *st);
2f6079dc 1306extern int fd2devnum(int fd);
b5e64645 1307
a931db9e
NB
1308static inline int dev2major(int d)
1309{
1310 if (d >= 0)
1311 return MD_MAJOR;
1312 else
1313 return get_mdp_major();
1314}
1315
1316static inline int dev2minor(int d)
1317{
1318 if (d >= 0)
1319 return d;
1320 return (-1-d) << MdpMinorShift;
1321}
1322
2d762ade 1323#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1324#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1325#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1326 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1327
3c558363
N
1328static inline int is_subarray(char *vers)
1329{
e9dd1598
N
1330 /* The version string for a 'subarray' (an array in a container)
1331 * is
1332 * /containername/componentname for normal read-write arrays
746a6567
N
1333 * -containername/componentname for arrays which mdmon must not
1334 * reconfigure. They might be read-only
1335 * or might be undergoing reshape etc.
e9dd1598
N
1336 * containername is e.g. md0, md_d1
1337 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1338 */
1339 return (*vers == '/' || *vers == '-');
3c558363
N
1340}
1341
e5408a32
DW
1342static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1343{
1344 return &ent->metadata_version[10+strlen(container)+1];
1345}
1346
2a24d7b6
DW
1347#ifdef DEBUG
1348#define dprintf(fmt, arg...) \
1349 fprintf(stderr, fmt, ##arg)
1350#else
1351#define dprintf(fmt, arg...) \
1352 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1353#endif
1a0ee0ba
DK
1354#include <assert.h>
1355#include <stdarg.h>
1356static inline int xasprintf(char **strp, const char *fmt, ...) {
1357 va_list ap;
1358 int ret;
1359 va_start(ap, fmt);
089485cb 1360 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1361 va_end(ap);
1362 assert(ret >= 0);
1363 return ret;
1364}
2a24d7b6 1365
e7b84f9d
N
1366#define pr_err(fmt ...) fprintf(stderr, Name ": " fmt)
1367#define cont_err(fmt ...) fprintf(stderr, " " fmt)
1368
503975b9
N
1369void *xmalloc(size_t len);
1370void *xrealloc(void *ptr, size_t len);
1371void *xcalloc(size_t num, size_t size);
1372char *xstrdup(const char *str);
1373
b5e64645
NB
1374#define LEVEL_MULTIPATH (-4)
1375#define LEVEL_LINEAR (-1)
1376#define LEVEL_FAULTY (-5)
1377
17f25ca6
NB
1378/* kernel module doesn't know about these */
1379#define LEVEL_CONTAINER (-100)
a322f70c 1380#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1381
b5e64645
NB
1382
1383/* faulty stuff */
1384
1385#define WriteTransient 0
1386#define ReadTransient 1
1387#define WritePersistent 2
1388#define ReadPersistent 3
1389#define WriteAll 4 /* doesn't go to device */
1390#define ReadFixable 5
1391#define Modes 6
1392
1393#define ClearErrors 31
1394#define ClearFaults 30
1395
1396#define AllPersist 100 /* internal use only */
1397#define NoPersist 101
1398
1399#define ModeMask 0x1f
1400#define ModeShift 5
fe6729fa
NB
1401
1402
1403#ifdef __TINYC__
1404#undef minor
1405#undef major
1406#undef makedev
1407#define minor(x) ((x)&0xff)
1408#define major(x) (((x)>>8)&0xff)
1409#define makedev(M,m) (((M)<<8) | (m))
1410#endif
1411
b640a252 1412/* for raid4/5/6 */
e86c9dd6
NB
1413#define ALGORITHM_LEFT_ASYMMETRIC 0
1414#define ALGORITHM_RIGHT_ASYMMETRIC 1
1415#define ALGORITHM_LEFT_SYMMETRIC 2
1416#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1417
1418/* Define non-rotating (raid4) algorithms. These allow
1419 * conversion of raid4 to raid5.
1420 */
1421#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1422#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1423
1424/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1425 * Firstly, the exact positioning of the parity block is slightly
1426 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1427 * of DDF.
1428 * Secondly, or order of datablocks over which the Q syndrome is computed
1429 * is different.
1430 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1431 * These layouts are from the DDFv1.2 spec.
1432 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1433 * leaves RLQ=3 as 'Vendor Specific'
1434 */
1435
1436#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1437#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1438#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1439
1440
1441/* For every RAID5 algorithm we define a RAID6 algorithm
1442 * with exactly the same layout for data and parity, and
1443 * with the Q block always on the last device (N-1).
1444 * This allows trivial conversion from RAID5 to RAID6
1445 */
1446#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1447#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1448#define ALGORITHM_LEFT_SYMMETRIC_6 18
1449#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1450#define ALGORITHM_PARITY_0_6 20
1451#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1452
33a6535d
AW
1453/* Define PATH_MAX in case we don't use glibc or standard library does
1454 * not have PATH_MAX defined. Assume max path length is 4K characters.
1455 */
1456#ifndef PATH_MAX
1457#define PATH_MAX 4096
1458#endif
1459
9dad51d4
N
1460#define RESYNC_NONE -1
1461#define RESYNC_DELAYED -2
1462#define RESYNC_PENDING -3
1463#define RESYNC_UNKNOWN -4
a0963a86 1464
480f3566
N
1465/* When using "GET_DISK_INFO" it isn't certain how high
1466 * we need to check. So we impose an absolute limit of
1467 * MAX_DISKS. This needs to be much more than the largest
1468 * number of devices any metadata can support. Currently
1469 * v1.x can support 1920
1470 */
1471#define MAX_DISKS 4096
1472
d04f65f4
N
1473/* Sometimes the 'size' value passed needs to mean "Maximum".
1474 * In those cases with use MAX_SIZE
1475 */
1476#define MAX_SIZE 1
822e393a
N
1477
1478/* We want to use unsigned numbers for sector counts, but need
1479 * a value for 'invalid'. Use '1'.
1480 */
1481#define INVALID_SECTORS 1
1482
a0963a86 1483extern int __offroot;