]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
super1: use awrite when writing a new bitmap.
[thirdparty/mdadm.git] / mdadm.h
CommitLineData
64c4757e 1/*
9a9dab36 2 * mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e 3 *
e736b623 4 * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
64c4757e
NB
5 *
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 * Author: Neil Brown
e736b623 22 * Email: <neilb@suse.de>
64c4757e
NB
23 */
24
4ccdb956 25#define _GNU_SOURCE
c2c9bb6f 26#define _FILE_OFFSET_BITS 64
64c4757e 27#include <unistd.h>
e39b673e 28#if !defined(__dietlibc__) && !defined(__KLIBC__)
64c4757e 29extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
f783ca4f 30#else
fffdbe5e 31# if defined(__NO_STAT64) || __WORDSIZE != 32
f783ca4f
NB
32# define lseek64 lseek
33# endif
98c6faba 34#endif
64c4757e
NB
35
36#include <sys/types.h>
37#include <sys/stat.h>
38#include <stdlib.h>
39#include <time.h>
11018a45 40#include <sys/time.h>
64c4757e
NB
41#include <getopt.h>
42#include <fcntl.h>
43#include <stdio.h>
44#include <errno.h>
45#include <string.h>
773135f5 46#include <syslog.h>
b56c3630
NB
47#ifdef __dietlibc__
48#include <strings.h>
280a927d
NB
49/* dietlibc has deprecated random and srandom!! */
50#define random rand
51#define srandom srand
98c6faba
NB
52#endif
53
64c4757e
NB
54
55#include <linux/kdev_t.h>
e0d19036
NB
56/*#include <linux/fs.h> */
57#include <sys/mount.h>
58#include <asm/types.h>
64c4757e
NB
59#include <sys/ioctl.h>
60#define MD_MAJOR 9
dd0781e5 61#define MdpMinorShift 6
64c4757e 62
e0d19036 63#ifndef BLKGETSIZE64
98c6faba 64#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 65#endif
56eb10c0 66
c21e737b 67#define DEFAULT_CHUNK 512
c82f047c
NB
68#define DEFAULT_BITMAP_CHUNK 4096
69#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 70#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e 71
753cf905 72/* MAP_DIR should be somewhere that persists across the pivotroot
5d4d1b26 73 * from early boot to late boot.
7f5de63d 74 * Currently /dev seems to be the only option on most distros.
5d4d1b26 75 */
753cf905 76#ifndef MAP_DIR
7f5de63d 77#define MAP_DIR "/dev/.mdadm"
753cf905
DL
78#endif /* MAP_DIR */
79/* MAP_FILE is what we name the map file we put in MAP_DIR, in case you
80 * want something other than the default of "map"
81 */
82#ifndef MAP_FILE
83#define MAP_FILE "map"
84#endif /* MAP_FILE */
85/* MDMON_DIR is where pid and socket files used for communicating
86 * with mdmon normally live. It *should* be /var/run, but when
87 * mdmon is needed at early boot then it needs to write there prior
88 * to /var/run being mounted read/write, and it also then needs to
89 * persist beyond when /var/run is mounter read-only. So, to be
90 * safe, the default is somewhere that is read/write early in the
91 * boot process and stays up as long as possible during shutdown.
92 */
93#ifndef MDMON_DIR
7f5de63d 94#define MDMON_DIR "/dev/.mdadm/"
753cf905 95#endif /* MDMON_DIR */
5d4d1b26 96
403410eb
PC
97/* FAILED_SLOTS is where to save files storing recent removal of array
98 * member in order to allow future reuse of disk inserted in the same
99 * slot for array recovery
100 */
101#ifndef FAILED_SLOTS_DIR
102#define FAILED_SLOTS_DIR "/dev/.mdadm/failed-slots"
103#endif /* FAILED_SLOTS */
104
64c4757e 105#include "md_u.h"
e0d19036 106#include "md_p.h"
c82f047c 107#include "bitmap.h"
f7dd881f 108#include "msg.h"
64c4757e 109
1e0d770c 110#include <endian.h>
efd441d1
NB
111/* Redhat don't like to #include <asm/byteorder.h>, and
112 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
113 * and there is no standard conversion function so... */
0ae03b8a
NB
114/* And dietlibc doesn't think byteswap is ok, so.. */
115/* #include <byteswap.h> */
116#define bswap_16(x) (((x) & 0x00ffU) << 8 | \
117 ((x) & 0xff00U) >> 8)
118#define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
119 ((x) & 0xff000000U) >> 24 | \
120 ((x) & 0x0000ff00U) << 8 | \
121 ((x) & 0x00ff0000U) >> 8)
122#define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
123 ((x) & 0xff00000000000000ULL) >> 56 | \
124 ((x) & 0x000000000000ff00ULL) << 40 | \
125 ((x) & 0x00ff000000000000ULL) >> 40 | \
126 ((x) & 0x0000000000ff0000ULL) << 24 | \
127 ((x) & 0x0000ff0000000000ULL) >> 24 | \
128 ((x) & 0x00000000ff000000ULL) << 8 | \
129 ((x) & 0x000000ff00000000ULL) >> 8)
130
e39b673e 131#if !defined(__KLIBC__)
efd441d1
NB
132#if BYTE_ORDER == LITTLE_ENDIAN
133#define __cpu_to_le16(_x) (_x)
134#define __cpu_to_le32(_x) (_x)
135#define __cpu_to_le64(_x) (_x)
136#define __le16_to_cpu(_x) (_x)
137#define __le32_to_cpu(_x) (_x)
138#define __le64_to_cpu(_x) (_x)
974e620d
NB
139
140#define __cpu_to_be16(_x) bswap_16(_x)
141#define __cpu_to_be32(_x) bswap_32(_x)
142#define __cpu_to_be64(_x) bswap_64(_x)
143#define __be16_to_cpu(_x) bswap_16(_x)
144#define __be32_to_cpu(_x) bswap_32(_x)
145#define __be64_to_cpu(_x) bswap_64(_x)
efd441d1
NB
146#elif BYTE_ORDER == BIG_ENDIAN
147#define __cpu_to_le16(_x) bswap_16(_x)
148#define __cpu_to_le32(_x) bswap_32(_x)
149#define __cpu_to_le64(_x) bswap_64(_x)
150#define __le16_to_cpu(_x) bswap_16(_x)
151#define __le32_to_cpu(_x) bswap_32(_x)
152#define __le64_to_cpu(_x) bswap_64(_x)
974e620d
NB
153
154#define __cpu_to_be16(_x) (_x)
155#define __cpu_to_be32(_x) (_x)
156#define __cpu_to_be64(_x) (_x)
157#define __be16_to_cpu(_x) (_x)
158#define __be32_to_cpu(_x) (_x)
159#define __be64_to_cpu(_x) (_x)
1e0d770c
NB
160#else
161# error "unknown endianness."
162#endif
e39b673e 163#endif /* __KLIBC__ */
1e0d770c
NB
164
165
1e5c6983
DW
166/*
167 * min()/max()/clamp() macros that also do
168 * strict type-checking.. See the
169 * "unnecessary" pointer comparison.
170 */
171#define min(x, y) ({ \
172 typeof(x) _min1 = (x); \
173 typeof(y) _min2 = (y); \
174 (void) (&_min1 == &_min2); \
175 _min1 < _min2 ? _min1 : _min2; })
176
177#define max(x, y) ({ \
178 typeof(x) _max1 = (x); \
179 typeof(y) _max2 = (y); \
180 (void) (&_max1 == &_max2); \
181 _max1 > _max2 ? _max1 : _max2; })
1e0d770c 182
4b1ac34b
NB
183/* general information that might be extracted from a superblock */
184struct mdinfo {
185 mdu_array_info_t array;
186 mdu_disk_info_t disk;
187 __u64 events;
fbf8a0b7 188 int uuid[4];
31317663 189 char name[33];
353632d9 190 unsigned long long data_offset;
598f0d58
NB
191 unsigned long long component_size; /* same as array.size, except in
192 * sectors and up to 64bits.
193 */
da9b4a62
DW
194 unsigned long long custom_array_size; /* size for non-default sized
195 * arrays (in sectors)
196 */
81219e70
LM
197#define NO_RESHAPE 0
198#define VOLUME_RESHAPE 1
199#define CONTAINER_RESHAPE 2
353632d9
NB
200 int reshape_active;
201 unsigned long long reshape_progress;
6e75048b
AK
202 int recovery_blocked; /* for external metadata it
203 * indicates that there is
204 * reshape in progress in
205 * container,
206 * for native metadata it is
207 * reshape_active field mirror
208 */
e1516be1
DW
209 union {
210 unsigned long long resync_start; /* per-array resync position */
211 unsigned long long recovery_start; /* per-device rebuild position */
b7528a20 212 #define MaxSector (~0ULL) /* resync/recovery complete position */
e1516be1 213 };
a67dd8cc 214 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 215 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 216 int errors;
f21e18ca 217 unsigned long cache_size; /* size of raid456 stripe cache*/
7e0f6979
NB
218 int mismatch_cnt;
219 char text_version[50];
dd15dc4a
NB
220
221 int container_member; /* for assembling external-metatdata arrays
222 * This is to be used internally by metadata
223 * handler only */
97b4d0e9
DW
224 int container_enough; /* flag external handlers can set to
225 * indicate that subarrays have not enough (-1),
226 * enough to start (0), or all expected disks (1) */
06c7f68e 227 char sys_name[20];
7e0f6979 228 struct mdinfo *devs;
06c7f68e 229 struct mdinfo *next;
549e9569
NB
230
231 /* Device info for mdmon: */
e1516be1 232 int recovery_fd;
549e9569 233 int state_fd;
8d45d196
DW
234 #define DS_FAULTY 1
235 #define DS_INSYNC 2
236 #define DS_WRITE_MOSTLY 4
237 #define DS_SPARE 8
238 #define DS_BLOCKED 16
239 #define DS_REMOVE 1024
92967543 240 #define DS_UNBLOCK 2048
549e9569
NB
241 int prev_state, curr_state, next_state;
242
4b1ac34b
NB
243};
244
5bbb4842
NB
245struct createinfo {
246 int uid;
247 int gid;
248 int autof;
249 int mode;
38098016 250 int symlinks;
058574b1 251 struct supertype *supertype;
5bbb4842
NB
252};
253
9a9dab36 254#define Name "mdadm"
682c7051 255
e0d19036
NB
256enum mode {
257 ASSEMBLE=1,
258 BUILD,
259 CREATE,
260 MANAGE,
261 MISC,
262 MONITOR,
dd0781e5 263 GROW,
8382f19b 264 INCREMENTAL,
1f48664b 265 AUTODETECT,
e0d19036
NB
266};
267
64c4757e 268extern char short_options[];
024768c4 269extern char short_bitmap_options[];
c06487ce 270extern char short_bitmap_auto_options[];
64c4757e 271extern struct option long_options[];
56eedc1a 272extern char Version[], Usage[], Help[], OptionHelp[],
dd0781e5 273 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 274 Help_incr[],
e0d19036 275 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 276
997aed5d 277/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 278 * numbers later than any 'short' character option.
997aed5d
NB
279 */
280enum special_options {
1c7a808c 281 AssumeClean = 300,
997aed5d
NB
282 BitmapChunk,
283 WriteBehind,
284 ReAdd,
285 NoDegraded,
286 Sparc22,
1c7a808c 287 BackupFile,
997aed5d 288 HomeHost,
589395d6 289 AutoHomeHost,
38098016 290 Symlinks,
1f48664b 291 AutoDetect,
1770662b 292 Waitclean,
4cce4069 293 DetailPlatform,
33414a01 294 KillSubarray,
1c7a808c 295 UpdateSubarray,
edde9560 296 IncrementalPath,
1c7a808c
N
297 NoSharing,
298 HelpOptions,
299 Brief,
300 ManageOpt,
301 Add,
302 Remove,
303 Fail,
304 MiscOpt,
305 WaitOpt,
306 ConfigFile,
307 ChunkSize,
308 WriteMostly,
309 Layout,
310 Auto,
311 Force,
312 SuperMinor,
313 EMail,
314 ProgramOpt,
315 Increment,
316 Fork,
317 Bitmap,
318 RebuildMapOpt,
87f26d14 319 InvalidBackup,
20b60dcd 320 UdevRules,
b76b30e0 321 FreezeReshape,
2dddadb0 322 Continue,
997aed5d
NB
323};
324
64c4757e 325/* structures read from config file */
52826846
NB
326/* List of mddevice names and identifiers
327 * Identifiers can be:
328 * uuid=128-hex-uuid
329 * super-minor=decimal-minor-number-from-superblock
330 * devices=comma,separated,list,of,device,names,with,wildcards
331 *
332 * If multiple fields are present, the intersection of all matching
333 * devices is considered
334 */
98c6faba 335#define UnSet (0xfffe)
fa56eddb 336struct mddev_ident {
dd0781e5 337 char *devname;
aba69144 338
dd0781e5 339 int uuid_set;
3fa06e9d 340 int uuid[4];
947fd4dd 341 char name[33];
52826846 342
f21e18ca 343 int super_minor;
52826846 344
dd0781e5 345 char *devices; /* comma separated list of device
52826846
NB
346 * names with wild cards
347 */
dd0781e5 348 int level;
f21e18ca
N
349 int raid_disks;
350 int spare_disks;
82d9eba6 351 struct supertype *st;
dd0781e5
NB
352 int autof; /* 1 for normal, 2 for partitioned */
353 char *spare_group;
7ef02d01 354 char *bitmap_file;
c82f047c 355 int bitmap_fd;
dd0781e5 356
1771a6e2
N
357 char *container; /* /dev/whatever name of container, or
358 * uuid of container. You would expect
359 * this to be the 'devname' or UUID
dbb44303
N
360 * of some other entry.
361 */
362 char *member; /* subarray within a container */
363
fa56eddb 364 struct mddev_ident *next;
b179246f
N
365 union {
366 /* fields needed by different users of this structure */
367 int assembled; /* set when assembly succeeds */
368 };
fa56eddb 369};
64c4757e
NB
370
371/* List of device names - wildcards expanded */
a655e550 372struct mddev_dev {
64c4757e 373 char *devname;
1c7a808c 374 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
cd29a5c8
NB
375 * Not set for names read from .config
376 */
b3d31955 377 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
fe80f49b 378 char re_add;
811e6cbe 379 char used; /* set when used */
a655e550
N
380 struct mddev_dev *next;
381};
64c4757e 382
682c7051
NB
383typedef struct mapping {
384 char *name;
385 int num;
386} mapping_t;
387
e0d19036
NB
388
389struct mdstat_ent {
390 char *dev;
391 int devnum;
392 int active;
393 char *level;
394 char *pattern; /* U or up, _ for down */
395 int percent; /* -1 if no resync */
f94c116f 396 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
397 int devcnt;
398 int raid_disks;
549e9569 399 char * metadata_version;
3b57c466
N
400 struct dev_member {
401 char *name;
402 struct dev_member *next;
403 } *members;
e0d19036
NB
404 struct mdstat_ent *next;
405};
406
22a88995 407extern struct mdstat_ent *mdstat_read(int hold, int start);
e0d19036 408extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 409extern void mdstat_wait(int seconds);
58a4ba2a 410extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
8382f19b 411extern int mddev_busy(int devnum);
3b57c466 412extern struct mdstat_ent *mdstat_by_component(char *name);
78b10e66 413extern struct mdstat_ent *mdstat_by_subdev(char *subdev, int container);
8382f19b
NB
414
415struct map_ent {
416 struct map_ent *next;
417 int devnum;
1522c538 418 char metadata[20];
8382f19b 419 int uuid[4];
195254b8 420 int bad;
8382f19b
NB
421 char *path;
422};
1522c538 423extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
8382f19b 424 int uuid[4], char *path);
4eb26970 425extern void map_remove(struct map_ent **map, int devnum);
8382f19b 426extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
f2e55ecc
N
427extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
428extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
429extern void map_read(struct map_ent **melp);
430extern int map_write(struct map_ent *mel);
431extern void map_delete(struct map_ent **mapp, int devnum);
432extern void map_free(struct map_ent *map);
433extern void map_add(struct map_ent **melp,
1522c538 434 int devnum, char *metadata, int uuid[4], char *path);
ad5bc697
N
435extern int map_lock(struct map_ent **melp);
436extern void map_unlock(struct map_ent **melp);
cc700db3 437extern void map_fork(void);
e0d19036 438
e86c9dd6 439/* various details can be requested */
dab4a513
DW
440enum sysfs_read_flags {
441 GET_LEVEL = (1 << 0),
442 GET_LAYOUT = (1 << 1),
443 GET_COMPONENT = (1 << 2),
444 GET_CHUNK = (1 << 3),
445 GET_CACHE = (1 << 4),
446 GET_MISMATCH = (1 << 5),
447 GET_VERSION = (1 << 6),
448 GET_DISKS = (1 << 7),
449 GET_DEGRADED = (1 << 8),
450 GET_SAFEMODE = (1 << 9),
451 GET_DEVS = (1 << 10), /* gets role, major, minor */
452 GET_OFFSET = (1 << 11),
453 GET_SIZE = (1 << 12),
454 GET_STATE = (1 << 13),
455 GET_ERROR = (1 << 14),
dab4a513 456};
e86c9dd6
NB
457
458/* If fd >= 0, get the array it is open on,
459 * else use devnum. >=0 -> major9. <0.....
460 */
549e9569 461extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 462extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
463extern void sysfs_free(struct mdinfo *sra);
464extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
465extern int sysfs_attr_match(const char *attr, const char *str);
466extern int sysfs_match_word(const char *word, char **list);
7e0f6979 467extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 468 char *name, char *val);
7e0f6979 469extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 470 char *name, unsigned long long val);
97590376 471extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
472extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
473 char *name);
474extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 475extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 476 char *name, unsigned long long *val);
7236ee7a 477extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
478extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
479 char *name);
93ecfa01
N
480extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
481 char *name, char *val, int size);
8ed3e5e1 482extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 483extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 484extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 485extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 486extern int sysfs_unique_holder(int devnum, long rdev);
bc77ed53 487extern int sysfs_freeze_array(struct mdinfo *sra);
c69b251b 488extern int load_sys(char *path, char *buf);
130994cb
AK
489extern int reshape_prepare_fdlist(char *devname,
490 struct mdinfo *sra,
491 int raid_disks,
492 int nrdisks,
493 unsigned long blocks,
494 char *backup_file,
495 int *fdlist,
496 unsigned long long *offsets);
497extern void reshape_free_fdlist(int *fdlist,
498 unsigned long long *offsets,
499 int size);
e6e9d47b
AK
500extern int reshape_open_backup_file(char *backup,
501 int fd,
502 char *devname,
503 long blocks,
504 int *fdlist,
a93f87ee
N
505 unsigned long long *offsets,
506 int restart);
1c009fc2
AK
507extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
508 unsigned int ndata, unsigned int odata);
e86c9dd6
NB
509
510extern int save_stripes(int *source, unsigned long long *offsets,
511 int raid_disks, int chunk_size, int level, int layout,
512 int nwrites, int *dest,
a6288483
N
513 unsigned long long start, unsigned long long length,
514 char *buf);
353632d9
NB
515extern int restore_stripes(int *dest, unsigned long long *offsets,
516 int raid_disks, int chunk_size, int level, int layout,
517 int source, unsigned long long read_offset,
2fcb75ae
AK
518 unsigned long long start, unsigned long long length,
519 char *src_buf);
e86c9dd6 520
52826846
NB
521#ifndef Sendmail
522#define Sendmail "/usr/lib/sendmail -t"
523#endif
524
773135f5
NB
525#define SYSLOG_FACILITY LOG_DAEMON
526
682c7051
NB
527extern char *map_num(mapping_t *map, int num);
528extern int map_name(mapping_t *map, char *name);
b640a252 529extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 530
16c6fa80 531extern char *map_dev(int major, int minor, int create);
64c4757e 532
549e9569 533struct active_array;
2e735d19 534struct metadata_update;
64c4757e 535
999b4972
N
536
537/* 'struct reshape' records the intermediate states
538 * a general reshape.
539 * The starting geometry is converted to the 'before' geometry
540 * by at most an atomic level change. They could be the same.
541 * Similarly the 'after' geometry is converted to the final
542 * geometry by at most a level change.
543 * Note that 'before' and 'after' must have the same level.
544 * 'blocks' is the minimum number of sectors for a reshape unit.
545 * This will be a multiple of the stripe size in each of the
546 * 'before' and 'after' geometries.
547 * If 'blocks' is 0, no restriping is necessary.
548 */
549struct reshape {
550 int level;
551 int parity; /* number of parity blocks/devices */
552 struct {
553 int layout;
554 int data_disks;
555 } before, after;
556 unsigned long long backup_blocks;
557 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
558 unsigned long long new_size; /* New size of array in sectors */
559};
560
6adfd3af
NB
561/* A superswitch provides entry point the a metadata handler.
562 *
563 * The super_switch primarily operates on some "metadata" that
564 * is accessed via the 'supertype'.
565 * This metadata has one of three possible sources.
566 * 1/ It is read from a single device. In this case it may not completely
567 * describe the array or arrays as some information might be on other
568 * devices.
569 * 2/ It is read from all devices in a container. In this case all
570 * information is present.
571 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
572 * be complete once enough ->add_to_super calls have completed.
573 *
574 * When creating an array inside a container, the metadata will be
575 * formed by a combination of 2 and 3. The metadata or the array is read,
576 * then new information is added.
577 *
578 * The metadata must sometimes have a concept of a 'current' array
579 * and a 'current' device.
580 * The 'current' array is set by init_super to be the newly created array,
581 * or is set by super_by_fd when it finds it is looking at an array inside
582 * a container.
583 *
584 * The 'current' device is either the device that the metadata was read from
585 * in case 1, or the last device added by add_to_super in case 3.
586 * Case 2 does not identify a 'current' device.
587 */
f9ce90ba 588extern struct superswitch {
6adfd3af
NB
589
590 /* Used to report details of metadata read from a component
591 * device. ->load_super has been called.
592 */
3da92f27 593 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 594 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 595 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 596 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
597
598 /* Used to report details of an active array.
599 * ->load_super was possibly given a 'component' string.
600 */
3da92f27 601 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 602 void (*brief_detail_super)(struct supertype *st);
0d726f17 603 void (*export_detail_super)(struct supertype *st);
6adfd3af 604
4cce4069 605 /* Optional: platform hardware / firmware details */
5615172f 606 int (*detail_platform)(int verbose, int enumerate_only);
4cce4069 607
6adfd3af
NB
608 /* Used:
609 * to get uuid to storing in bitmap metadata
610 * and 'reshape' backup-data metadata
611 * To see if a device is being re-added to an array it was part of.
612 */
3da92f27 613 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 614
8592f29d 615 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
616 * the container, or about an individual array within the container.
617 * The determination is made either by:
618 * load_super being given a 'component' string.
619 * validate_geometry determining what to create.
d2ca6449
NB
620 * The info includes both array information and device information.
621 * The particular device should be:
622 * The last device added by add_to_super
623 * The device the metadata was loaded from by load_super
a5d85af7
N
624 * If 'map' is present, then it is an array raid_disks long
625 * (raid_disk must already be set and correct) and it is filled
626 * with 1 for slots that are thought to be active and 0 for slots which
627 * appear to be failed/missing.
95eeceeb 628 * *info is zeroed out before data is added.
6adfd3af 629 */
a5d85af7 630 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 631 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 632 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 633 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 634 */
3da92f27 635 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
636
637 /* Make one of several generic modifications to metadata
638 * prior to assembly (or other times).
639 * sparc2.2 - first bug in early 0.90 metadata
640 * super-minor - change name of 0.90 metadata
641 * summaries - 'correct' any redundant data
642 * resync - mark array as dirty to trigger a resync.
643 * uuid - set new uuid - only 0.90 or 1.x
644 * name - change the name of the array (where supported)
645 * homehost - change which host this array is tied to.
646 * devicesize - If metadata is at start of device, change recorded
647 * device size to match actual device size
648 * byteorder - swap bytes for 0.90 metadata
649 *
650 * force-one - mark that device as uptodate, not old or failed.
651 * force-array - mark array as clean if it would not otherwise
652 * assemble
653 * assemble - not sure how this is different from force-one...
654 * linear-grow-new - add a new device to a linear array, but don't
655 * change the size: so superblock still matches
656 * linear-grow-update - now change the size of the array.
16715c01
DL
657 * writemostly - set the WriteMostly1 bit in the superblock devflags
658 * readwrite - clear the WriteMostly1 bit in the superblock devflags
6adfd3af 659 */
68c7d6d7 660 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 661 char *update,
e5eac01f
NB
662 char *devname, int verbose,
663 int uuid_set, char *homehost);
6adfd3af
NB
664
665 /* Create new metadata for new array as described. This could
666 * be a new container, or an array in a pre-existing container.
667 * Also used to zero metadata prior to writing it to invalidate old
668 * metadata.
669 */
3da92f27
NB
670 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
671 unsigned long long size, char *name,
672 char *homehost, int *uuid);
6adfd3af
NB
673
674 /* update the metadata to include new device, either at create or
675 * when hot-adding a spare.
676 */
f20c3968 677 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
1a64be56
LM
678 int fd, char *devname);
679 /* update the metadata to delete a device,
680 * when hot-removing.
681 */
682 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
683
684 /* Write metadata to one device when fixing problems or adding
685 * a new device.
686 */
3da92f27 687 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
688
689 /* Write all metadata for this array.
690 */
111d01fc 691 int (*write_init_super)(struct supertype *st);
64557c33 692 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 693 int (*load_super)(struct supertype *st, int fd, char *devname);
2b959fbf 694 int (*load_container)(struct supertype *st, int fd, char *devname);
82d9eba6 695 struct supertype * (*match_metadata_desc)(char *arg);
1bf4e2d9 696 __u64 (*avail_size)(struct supertype *st, __u64 size);
80e7f8c3 697 unsigned long long (*min_acceptable_spare_size)(struct supertype *st);
3da92f27 698 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 699 int delay, int write_behind,
f9c25f1d 700 unsigned long long size, int may_change, int major);
3da92f27
NB
701 void (*locate_bitmap)(struct supertype *st, int fd);
702 int (*write_bitmap)(struct supertype *st, int fd);
703 void (*free_super)(struct supertype *st);
78e44928
NB
704
705 /* validate_geometry is called with an st returned by
706 * match_metadata_desc.
707 * It should check that the geometry described in compatible with
708 * the metadata type. It will be called repeatedly as devices
709 * added to validate changing size and new devices. If there are
710 * inter-device dependencies, it should record sufficient details
711 * so these can be validated.
30f58b22 712 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
713 * Return value is:
714 * 1: everything is OK
715 * 0: not OK for some reason - if 'verbose', then error was reported.
716 * -1: st->sb was NULL, 'subdev' is a member of a container of this
717 * types, but array is not acceptable for some reason
718 * message was reported even if verbose is 0.
78e44928 719 */
17f25ca6
NB
720 int (*validate_geometry)(struct supertype *st, int level, int layout,
721 int raiddisks,
c21e737b 722 int *chunk, unsigned long long size,
2c514b71
NB
723 char *subdev, unsigned long long *freesize,
724 int verbose);
598f0d58 725
00bbdbda 726 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
727 /* query the supertype for default geometry */
728 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01
DW
729 /* Permit subarray's to be deleted from inactive containers */
730 int (*kill_subarray)(struct supertype *st); /* optional */
aa534678 731 /* Permit subarray's to be modified */
a951a4f7 732 int (*update_subarray)(struct supertype *st, char *subarray,
fa56eddb 733 char *update, struct mddev_ident *ident); /* optional */
7bc71196
DW
734 /* Check if reshape is supported for this external format.
735 * st is obtained from super_by_fd() where st->subarray[0] is
736 * initialized to indicate if reshape is being performed at the
737 * container or subarray level
738 */
739 int (*reshape_super)(struct supertype *st, long long size, int level,
740 int layout, int chunksize, int raid_disks,
41784c88
AK
741 int delta_disks, char *backup, char *dev,
742 int verbose); /* optional */
999b4972
N
743 int (*manage_reshape)( /* optional */
744 int afd, struct mdinfo *sra, struct reshape *reshape,
745 struct supertype *st, unsigned long blocks,
746 int *fds, unsigned long long *offsets,
747 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 748
549e9569 749/* for mdmon */
cba0191b
NB
750 int (*open_new)(struct supertype *c, struct active_array *a,
751 char *inst);
ed9d66aa
NB
752
753 /* Tell the metadata handler the current state of the array.
754 * This covers whether it is known to be consistent (no pending writes)
01f157d7 755 * and how far along a resync is known to have progressed
ed9d66aa
NB
756 * (in a->resync_start).
757 * resync status is really irrelevant if the array is not consistent,
758 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
759 * If 'consistent' is '2', then the array can mark it dirty if a
760 * resync/recovery/whatever is required, or leave it clean if not.
761 * Return value is 0 dirty (not consistent) and 1 if clean.
762 * it is only really important if consistent is passed in as '2'.
4e5528c6 763 */
01f157d7 764 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
765
766 /* When the state of a device might have changed, we call set_disk to
767 * tell the metadata what the current state is.
768 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
769 * transitions.
770 * set_disk might be called when the state of the particular disk has
771 * not in fact changed.
772 */
8d45d196 773 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
774 void (*sync_metadata)(struct supertype *st);
775 void (*process_update)(struct supertype *st,
776 struct metadata_update *update);
edd8d13c
NB
777 void (*prepare_update)(struct supertype *st,
778 struct metadata_update *update);
549e9569 779
6c3fb95c
NB
780 /* activate_spare will check if the array is degraded and, if it
781 * is, try to find some spare space in the container.
782 * On success, it add appropriate updates (For process_update) to
783 * to the 'updates' list and returns a list of 'mdinfo' identifying
784 * the device, or devices as there might be multiple missing
785 * devices and multiple spares available.
786 */
787 struct mdinfo *(*activate_spare)(struct active_array *a,
788 struct metadata_update **updates);
2cda7640
ML
789 /*
790 * Return statically allocated string that represents metadata specific
791 * controller domain of the disk. The domain is used in disk domain
792 * matching functions. Disks belong to the same domain if the they have
793 * the same domain from mdadm.conf and belong the same metadata domain.
794 * Returning NULL or not providing this handler means that metadata
795 * does not distinguish the differences between disks that belong to
796 * different controllers. They are in the domain specified by
797 * configuration file (mdadm.conf).
798 * In case when the metadata has the notion of domains based on disk
799 * it shall return NULL for disks that do not belong to the controller
800 * the supported domains. Such disks will form another domain and won't
801 * be mixed with supported ones.
802 */
803 const char *(*get_disk_controller_domain)(const char *path);
549e9569 804
ba53ea59
AK
805 /* for external backup area */
806 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
807
f277ce36 808 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 809 int external;
4cce4069 810 const char *name; /* canonical metadata name */
0f22b998 811} *superlist[];
82d9eba6 812
0f22b998
N
813extern struct superswitch super0, super1;
814extern struct superswitch super_imsm, super_ddf;
0592faeb 815extern struct superswitch mbr, gpt;
cdddbdbc 816
edd8d13c
NB
817struct metadata_update {
818 int len;
819 char *buf;
820 void *space; /* allocated space that monitor will use */
cb23f1f4
N
821 void **space_list; /* list of allocated spaces that monitor can
822 * use or that it returned.
823 */
edd8d13c
NB
824 struct metadata_update *next;
825};
826
6adfd3af
NB
827/* A supertype holds a particular collection of metadata.
828 * It identifies the metadata type by the superswitch, and the particular
829 * sub-version of that metadata type.
830 * metadata read in or created is stored in 'sb' and 'info'.
831 * There are also fields used by mdmon to track containers.
832 *
352452c3
N
833 * A supertype may refer to:
834 * Just an array, possibly in a container
835 * A container, not identifying any particular array
836 * Info read from just one device, not yet fully describing the array/container.
837 *
838 *
6adfd3af
NB
839 * A supertype is created by:
840 * super_by_fd
841 * guess_super
842 * dup_super
843 */
82d9eba6
NB
844struct supertype {
845 struct superswitch *ss;
846 int minor_version;
ea329559 847 int max_devs;
2f6079dc 848 int container_dev; /* devnum of container */
64557c33 849 void *sb;
111d01fc 850 void *info;
df3346e6
LM
851 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
852 HW/firmware related incompatability to load metadata.
853 Used when examining metadata to display content of disk
854 when user has no hw/firmare compatible system.
855 */
edd8d13c
NB
856 struct metadata_update *updates;
857 struct metadata_update **update_tail;
858
549e9569
NB
859 /* extra stuff used by mdmon */
860 struct active_array *arrays;
549e9569 861 int sock; /* listen to external programs */
549e9569
NB
862 int devnum;
863 char *devname; /* e.g. md0. This appears in metadata_verison:
864 * external:/md0/12
865 */
866 int devcnt;
867
868 struct mdinfo *devs;
869
82d9eba6 870};
f9ce90ba 871
4725bc31 872extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
873enum guess_types { guess_any, guess_array, guess_partitions };
874extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
875static inline struct supertype *guess_super(int fd) {
876 return guess_super_type(fd, guess_any);
877}
3da92f27 878extern struct supertype *dup_super(struct supertype *st);
beae1dfe 879extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
3a371610 880extern int must_be_container(int fd);
0f0749ad 881extern int dev_size_from_id(dev_t id, unsigned long long *size);
8382f19b
NB
882extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
883 mdu_disk_info_t *disk);
a7c6e3fb 884void wait_for(char *dev, int fd);
f9ce90ba 885
5527fc74
N
886/*
887 * Data structures for policy management.
888 * Each device can have a policy structure that lists
889 * various name/value pairs each possibly with a metadata associated.
890 * The policy list is sorted by name/value/metadata
891 */
892struct dev_policy {
893 struct dev_policy *next;
894 char *name; /* None of these strings are allocated. They are
895 * all just references to strings which are known
896 * to exist elsewhere.
897 * name and metadata can be compared by address equality.
898 */
899 const char *metadata;
2cda7640 900 const char *value;
5527fc74
N
901};
902
4e8d9f0a 903extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
904
905/* iterate over the sublist starting at list, having the same
906 * 'name' as 'list', and matching the given metadata (Where
907 * NULL matches anything
908 */
909#define pol_for_each(item, list, _metadata) \
910 for (item = list; \
911 item && item->name == list->name; \
912 item = item->next) \
913 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
914 ; else
915
916/*
917 * policy records read from mdadm are largely just name-value pairs.
918 * The names are constants, not strdupped
919 */
920struct pol_rule {
921 struct pol_rule *next;
922 char *type; /* rule_policy or rule_part */
923 struct rule {
924 struct rule *next;
925 char *name;
926 char *value;
927 char *dups; /* duplicates of 'value' with a partNN appended */
928 } *rule;
929};
930
931extern char rule_policy[], rule_part[];
932extern char rule_path[], rule_type[];
933extern char type_part[], type_disk[];
934
935extern void policyline(char *line, char *type);
4e8d9f0a 936extern void policy_add(char *type, ...);
5527fc74
N
937extern void policy_free(void);
938
73c9c47c 939extern struct dev_policy *path_policy(char *path, char *type);
5527fc74 940extern struct dev_policy *disk_policy(struct mdinfo *disk);
4e8d9f0a 941extern struct dev_policy *devnum_policy(int dev);
5527fc74
N
942extern void dev_policy_free(struct dev_policy *p);
943
e78dda3b
N
944//extern void pol_new(struct dev_policy **pol, char *name, char *val, char *metadata);
945extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
946extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
947
e3bb5f14
N
948enum policy_action {
949 act_default,
950 act_include,
951 act_re_add,
d2db3045
N
952 act_spare, /* This only applies to bare devices */
953 act_spare_same_slot, /* this allows non-bare devices,
954 * but only if recent removal */
955 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
956 act_err
957};
958
959extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
960 enum policy_action want);
961extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
962 enum policy_action want);
f5f12c84
N
963
964struct domainlist {
965 struct domainlist *next;
2cda7640 966 const char *dom;
f5f12c84
N
967};
968
969extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
970 const char *metadata);
971extern struct domainlist *domain_from_array(struct mdinfo *mdi,
972 const char *metadata);
e78dda3b
N
973extern void domainlist_add_dev(struct domainlist **dom, int devnum,
974 const char *metadata);
f5f12c84
N
975extern void domain_free(struct domainlist *dl);
976extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
977 const char *metadata);
e78dda3b 978void domain_add(struct domainlist **domp, char *domain);
f5f12c84 979
403410eb
PC
980extern void policy_save_path(char *id_path, struct map_ent *array);
981extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
982
838acbc2
NB
983#if __GNUC__ < 3
984struct stat64;
985#endif
986
45e878bb
NB
987#define HAVE_NFTW we assume
988#define HAVE_FTW
989
2df1f269 990#ifdef __UCLIBC__
45e878bb 991# include <features.h>
2df1f269
BRF
992# ifndef __UCLIBC_HAS_LFS__
993# define lseek64 lseek
994# endif
45e878bb
NB
995# ifndef __UCLIBC_HAS_FTW__
996# undef HAVE_FTW
997# undef HAVE_NFTW
998# endif
999#endif
1000
1001#ifdef __dietlibc__
1002# undef HAVE_NFTW
1003#endif
1004
e39b673e 1005#if defined(__KLIBC__)
1006# undef HAVE_NFTW
1007# undef HAVE_FTW
1008#endif
1009
45e878bb 1010#ifndef HAVE_NFTW
173fc515 1011# define FTW_PHYS 1
45e878bb
NB
1012# ifndef HAVE_FTW
1013 struct FTW {};
173fc515
NB
1014# endif
1015#endif
1016
45e878bb
NB
1017#ifdef HAVE_FTW
1018# include <ftw.h>
1019#endif
1020
173fc515
NB
1021extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1022
f9ce90ba 1023
64c4757e 1024extern int Manage_ro(char *devname, int fd, int readonly);
91f068bf 1025extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
dd0781e5 1026extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
64c4757e 1027extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1028 struct mddev_dev *devlist, int verbose, int test,
11b391ec 1029 char *update, int force);
1f48664b 1030extern int autodetect(void);
e5329c37 1031extern int Grow_Add_device(char *devname, int fd, char *newdev);
8fac0577 1032extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
06b0d786 1033extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
e86c9dd6 1034 long long size,
691a36b7 1035 int level, char *layout_str, int chunksize, int raid_disks,
e2e53a2d 1036 struct mddev_dev *devlist,
ce52f92f 1037 int assume_clean, int force);
06b0d786 1038extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1039 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1040extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0
AK
1041 struct mdinfo *info, char *backup_file,
1042 int freeze_reshape);
1043
3f54bd62
AK
1044extern int restore_backup(struct supertype *st,
1045 struct mdinfo *content,
1046 int working_disks,
1047 int spares,
1048 char *backup_file,
1049 int verbose);
2dddadb0
AK
1050extern int Grow_continue_command(char *devname, int fd,
1051 char *backup_file, int verbose);
64c4757e 1052
7f91af49 1053extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1054 struct mddev_ident *ident,
87f26d14
N
1055 struct mddev_dev *devlist,
1056 char *backup_file, int invalid_backup,
64c4757e 1057 int readonly, int runstop,
0ac91628 1058 char *update, char *homehost, int require_homehost,
b76b30e0 1059 int verbose, int force, int freeze_reshape);
64c4757e 1060
7f91af49 1061extern int Build(char *mddev, int chunk, int level, int layout,
a655e550 1062 int raiddisks, struct mddev_dev *devlist, int assume_clean,
7f91af49 1063 char *bitmap_file, int bitmap_chunk, int write_behind,
83208785 1064 int delay, int verbose, int autof, unsigned long long size);
64c4757e
NB
1065
1066
7f91af49 1067extern int Create(struct supertype *st, char *mddev,
5dd497ee 1068 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
3d3dd91e 1069 char *name, char *homehost, int *uuid,
a655e550 1070 int subdevs, struct mddev_dev *devlist,
47d79ef8 1071 int runstop, int verbose, int force, int assume_clean,
7f91af49 1072 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
64c4757e 1073
54bad364 1074extern int Detail(char *dev, int brief, int export, int test, char *homehost);
4cce4069 1075extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
e0d19036 1076extern int Query(char *dev);
a655e550 1077extern int Examine(struct mddev_dev *devlist, int brief, int export, int scan,
0d726f17 1078 int SparcAdjust, struct supertype *forcest, char *homehost);
a655e550 1079extern int Monitor(struct mddev_dev *devlist,
e0d19036 1080 char *mailaddr, char *alert_cmd,
aa88f531 1081 int period, int daemonise, int scan, int oneshot,
edde9560
AC
1082 int dosyslog, int test, char *pidfile, int increments,
1083 int share);
64c4757e 1084
9277cc77 1085extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
33414a01 1086extern int Kill_subarray(char *dev, char *subarray, int quiet);
fa56eddb 1087extern int Update_subarray(char *dev, char *subarray, char *update, struct mddev_ident *ident, int quiet);
b90c0e9a 1088extern int Wait(char *dev);
9f1da824 1089extern int WaitClean(char *dev, int sock, int verbose);
9a9dab36 1090
8382f19b 1091extern int Incremental(char *devname, int verbose, int runstop,
0ac91628 1092 struct supertype *st, char *homehost, int require_homehost,
b76b30e0 1093 int autof, int freeze_reshape);
8382f19b
NB
1094extern void RebuildMap(void);
1095extern int IncrementalScan(int verbose);
950bc344 1096extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1097extern int CreateBitmap(char *filename, int force, char uuid[16],
1098 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1099 unsigned long write_behind,
dcec9ee5
NB
1100 unsigned long long array_size,
1101 int major);
55935d51 1102extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
20b60dcd 1103extern int Write_rules(char *rule_name);
bf4fb153 1104extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 1105extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 1106
64c4757e 1107extern int md_get_version(int fd);
e0d19036 1108extern int get_linux_version(void);
bc77ed53 1109extern int mdadm_version(char *version);
84e11361 1110extern long long parse_size(char *size);
64c4757e 1111extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
1112extern int parse_layout_10(char *layout);
1113extern int parse_layout_faulty(char *layout);
682c7051
NB
1114extern int check_ext2(int fd, char *name);
1115extern int check_reiser(int fd, char *name);
1116extern int check_raid(int fd, char *name);
53ed6ac3
KW
1117extern int check_partitions(int fd, char *dname,
1118 unsigned long long freesize,
1119 unsigned long long size);
64c4757e 1120
dd0781e5 1121extern int get_mdp_major(void);
8b0dabea 1122extern int dev_open(char *dev, int flags);
e8a70c89 1123extern int open_dev(int devnum);
d998b738 1124extern int open_dev_flags(int devnum, int flags);
a931db9e 1125extern int open_dev_excl(int devnum);
8d80900b 1126extern int is_standard(char *dev, int *nump);
9008ed1c 1127extern int same_dev(char *one, char *two);
dd0781e5 1128
f1ae21c4 1129extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1130extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1131extern struct mddev_dev *conf_get_devs(void);
8382f19b 1132extern int conf_test_dev(char *devname);
4e8d9f0a 1133extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1134extern struct createinfo *conf_get_create_info(void);
1135extern void set_conffile(char *file);
1136extern char *conf_get_mailaddr(void);
1137extern char *conf_get_mailfrom(void);
1138extern char *conf_get_program(void);
0ac91628 1139extern char *conf_get_homehost(int *require_homehostp);
e0d19036 1140extern char *conf_line(FILE *file);
98c6faba 1141extern char *conf_word(FILE *file, int allow_key);
0ac91628 1142extern int conf_name_is_free(char *name);
7c336758 1143extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1144extern int devname_matches(char *name, char *match);
2244d1a9
N
1145extern struct mddev_ident *conf_match(struct supertype *st,
1146 struct mdinfo *info,
1147 char *devname,
1148 int verbose, int *rvp);
6d11ec6f 1149extern int experimental(void);
0ac91628 1150
e0d19036
NB
1151extern void free_line(char *line);
1152extern int match_oneof(char *devices, char *devname);
e0d19036 1153extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1154extern const int uuid_zero[4];
f277ce36 1155extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1156extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1157extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1158extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1159 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1160extern unsigned long calc_csum(void *super, int bytes);
583315d9 1161extern int enough(int level, int raid_disks, int layout, int clean,
265e0f17 1162 char *avail, int avail_disks);
8453e704 1163extern int enough_fd(int fd);
e0d19036 1164extern int ask(char *mesg);
8fac0577 1165extern unsigned long long get_component_size(int fd);
0430ed48 1166extern void remove_partitions(int fd);
691c6ee1 1167extern int test_partition(int fd);
bfd76b93 1168extern int test_partition_from_id(dev_t id);
577e8448 1169extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1170extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1171 int chunksize, unsigned long long devsize);
edd8d13c
NB
1172extern int flush_metadata_updates(struct supertype *st);
1173extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c
DW
1174extern int assemble_container_content(struct supertype *st, int mdfd,
1175 struct mdinfo *content, int runstop,
49680258 1176 char *chosen_name, int verbose,
b76b30e0 1177 char *backup_file, int freeze_reshape);
326727d9
AC
1178extern struct mdinfo *container_choose_spares(struct supertype *st,
1179 unsigned long long min_size,
1180 struct domainlist *domlist,
1181 char *spare_group,
1182 const char *metadata, int get_one);
d52bb542 1183extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1184extern int add_disk(int mdfd, struct supertype *st,
1185 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1186extern int remove_disk(int mdfd, struct supertype *st,
1187 struct mdinfo *sra, struct mdinfo *info);
f35f2525 1188extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1189unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1190
56eb10c0 1191extern char *human_size(long long bytes);
e4965ef8
N
1192extern char *human_size_brief(long long bytes);
1193extern void print_r10_layout(int layout);
e0d19036 1194
63152c1b
NB
1195#define NoMdDev (1<<23)
1196extern int find_free_devnum(int use_partitions);
1197
e0d19036
NB
1198extern void put_md_name(char *name);
1199extern char *get_md_name(int dev);
1200
c913b90e 1201extern char DefaultConfFile[];
b5e64645 1202
69207ff6
N
1203extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
1204 char *chosen);
1205/* values for 'trustworthy' */
1206#define LOCAL 1
d1d3482b 1207#define LOCAL_ANY 10
69207ff6
N
1208#define FOREIGN 2
1209#define METADATA 3
6be1d39d 1210extern int open_mddev(char *dev, int report_errors);
a322f70c 1211extern int open_container(int fd);
33414a01
DW
1212extern int is_container_member(struct mdstat_ent *ent, char *devname);
1213extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1214extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1215extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1216
a931db9e 1217extern int mdmon_running(int devnum);
24f6f99b 1218extern int mdmon_pid(int devnum);
40ebbb9c 1219extern int check_env(char *name);
148acb7b 1220extern __u32 random32(void);
8850ee3e 1221extern int start_mdmon(int devnum);
a931db9e 1222
999b4972
N
1223extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1224 struct supertype *st, unsigned long stripes,
1225 int *fds, unsigned long long *offsets,
1226 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1227void abort_reshape(struct mdinfo *sra);
999b4972 1228
2f6079dc 1229extern char *devnum2devname(int num);
11877f4d 1230extern void fmt_devname(char *name, int num);
77472ff8 1231extern int devname2devnum(char *name);
c94709e8 1232extern int stat2devnum(struct stat *st);
2f6079dc 1233extern int fd2devnum(int fd);
b5e64645 1234
a931db9e
NB
1235static inline int dev2major(int d)
1236{
1237 if (d >= 0)
1238 return MD_MAJOR;
1239 else
1240 return get_mdp_major();
1241}
1242
1243static inline int dev2minor(int d)
1244{
1245 if (d >= 0)
1246 return d;
1247 return (-1-d) << MdpMinorShift;
1248}
1249
6416d527
NB
1250static inline int ROUND_UP(int a, int base)
1251{
1252 return ((a+base-1)/base)*base;
1253}
1254
3c558363
N
1255static inline int is_subarray(char *vers)
1256{
e9dd1598
N
1257 /* The version string for a 'subarray' (an array in a container)
1258 * is
1259 * /containername/componentname for normal read-write arrays
746a6567
N
1260 * -containername/componentname for arrays which mdmon must not
1261 * reconfigure. They might be read-only
1262 * or might be undergoing reshape etc.
e9dd1598
N
1263 * containername is e.g. md0, md_d1
1264 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1265 */
1266 return (*vers == '/' || *vers == '-');
3c558363
N
1267}
1268
e5408a32
DW
1269static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1270{
1271 return &ent->metadata_version[10+strlen(container)+1];
1272}
1273
2a24d7b6
DW
1274#ifdef DEBUG
1275#define dprintf(fmt, arg...) \
1276 fprintf(stderr, fmt, ##arg)
1277#else
1278#define dprintf(fmt, arg...) \
1279 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1280#endif
1a0ee0ba
DK
1281#include <assert.h>
1282#include <stdarg.h>
1283static inline int xasprintf(char **strp, const char *fmt, ...) {
1284 va_list ap;
1285 int ret;
1286 va_start(ap, fmt);
089485cb 1287 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1288 va_end(ap);
1289 assert(ret >= 0);
1290 return ret;
1291}
2a24d7b6 1292
b5e64645
NB
1293#define LEVEL_MULTIPATH (-4)
1294#define LEVEL_LINEAR (-1)
1295#define LEVEL_FAULTY (-5)
1296
17f25ca6
NB
1297/* kernel module doesn't know about these */
1298#define LEVEL_CONTAINER (-100)
a322f70c 1299#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1300
b5e64645
NB
1301
1302/* faulty stuff */
1303
1304#define WriteTransient 0
1305#define ReadTransient 1
1306#define WritePersistent 2
1307#define ReadPersistent 3
1308#define WriteAll 4 /* doesn't go to device */
1309#define ReadFixable 5
1310#define Modes 6
1311
1312#define ClearErrors 31
1313#define ClearFaults 30
1314
1315#define AllPersist 100 /* internal use only */
1316#define NoPersist 101
1317
1318#define ModeMask 0x1f
1319#define ModeShift 5
fe6729fa
NB
1320
1321
1322#ifdef __TINYC__
1323#undef minor
1324#undef major
1325#undef makedev
1326#define minor(x) ((x)&0xff)
1327#define major(x) (((x)>>8)&0xff)
1328#define makedev(M,m) (((M)<<8) | (m))
1329#endif
1330
b640a252 1331/* for raid4/5/6 */
e86c9dd6
NB
1332#define ALGORITHM_LEFT_ASYMMETRIC 0
1333#define ALGORITHM_RIGHT_ASYMMETRIC 1
1334#define ALGORITHM_LEFT_SYMMETRIC 2
1335#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1336
1337/* Define non-rotating (raid4) algorithms. These allow
1338 * conversion of raid4 to raid5.
1339 */
1340#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1341#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1342
1343/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1344 * Firstly, the exact positioning of the parity block is slightly
1345 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1346 * of DDF.
1347 * Secondly, or order of datablocks over which the Q syndrome is computed
1348 * is different.
1349 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1350 * These layouts are from the DDFv1.2 spec.
1351 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1352 * leaves RLQ=3 as 'Vendor Specific'
1353 */
1354
1355#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1356#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1357#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1358
1359
1360/* For every RAID5 algorithm we define a RAID6 algorithm
1361 * with exactly the same layout for data and parity, and
1362 * with the Q block always on the last device (N-1).
1363 * This allows trivial conversion from RAID5 to RAID6
1364 */
1365#define ALGORITHM_LEFT_ASYMMETRIC_6 16
1366#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1367#define ALGORITHM_LEFT_SYMMETRIC_6 18
1368#define ALGORITHM_RIGHT_SYMMETRIC_6 19
1369#define ALGORITHM_PARITY_0_6 20
1370#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1371
33a6535d
AW
1372/* Define PATH_MAX in case we don't use glibc or standard library does
1373 * not have PATH_MAX defined. Assume max path length is 4K characters.
1374 */
1375#ifndef PATH_MAX
1376#define PATH_MAX 4096
1377#endif
1378
2d3603ba
KW
1379#define PROCESS_DELAYED -2
1380#define PROCESS_PENDING -3