]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Handle extra 'grow' variations.
[thirdparty/mdadm.git] / mdadm.h
CommitLineData
64c4757e 1/*
9a9dab36 2 * mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e 3 *
4f589ad0 4 * Copyright (C) 2001-2006 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
22 * Email: <neilb@cse.unsw.edu.au>
23 * Paper: Neil Brown
24 * School of Computer Science and Engineering
25 * The University of New South Wales
26 * Sydney, 2052
27 * Australia
28 */
29
4ccdb956 30#define _GNU_SOURCE
c2c9bb6f 31#define _FILE_OFFSET_BITS 64
64c4757e 32#include <unistd.h>
e39b673e 33#if !defined(__dietlibc__) && !defined(__KLIBC__)
64c4757e 34extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
f783ca4f 35#else
fffdbe5e 36# if defined(__NO_STAT64) || __WORDSIZE != 32
f783ca4f
NB
37# define lseek64 lseek
38# endif
98c6faba 39#endif
64c4757e
NB
40
41#include <sys/types.h>
42#include <sys/stat.h>
43#include <stdlib.h>
44#include <time.h>
11018a45 45#include <sys/time.h>
64c4757e
NB
46#include <getopt.h>
47#include <fcntl.h>
48#include <stdio.h>
49#include <errno.h>
50#include <string.h>
773135f5 51#include <syslog.h>
b56c3630
NB
52#ifdef __dietlibc__
53#include <strings.h>
280a927d
NB
54/* dietlibc has deprecated random and srandom!! */
55#define random rand
56#define srandom srand
98c6faba
NB
57#endif
58
64c4757e
NB
59
60#include <linux/kdev_t.h>
e0d19036
NB
61/*#include <linux/fs.h> */
62#include <sys/mount.h>
63#include <asm/types.h>
64c4757e
NB
64#include <sys/ioctl.h>
65#define MD_MAJOR 9
dd0781e5 66#define MdpMinorShift 6
64c4757e 67
e0d19036 68#ifndef BLKGETSIZE64
98c6faba 69#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 70#endif
56eb10c0 71
c82f047c
NB
72#define DEFAULT_BITMAP_CHUNK 4096
73#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 74#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e
NB
75
76#include "md_u.h"
e0d19036 77#include "md_p.h"
c82f047c 78#include "bitmap.h"
f7dd881f 79#include "msg.h"
64c4757e 80
1e0d770c 81#include <endian.h>
efd441d1
NB
82/* Redhat don't like to #include <asm/byteorder.h>, and
83 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
84 * and there is no standard conversion function so... */
0ae03b8a
NB
85/* And dietlibc doesn't think byteswap is ok, so.. */
86/* #include <byteswap.h> */
87#define bswap_16(x) (((x) & 0x00ffU) << 8 | \
88 ((x) & 0xff00U) >> 8)
89#define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
90 ((x) & 0xff000000U) >> 24 | \
91 ((x) & 0x0000ff00U) << 8 | \
92 ((x) & 0x00ff0000U) >> 8)
93#define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
94 ((x) & 0xff00000000000000ULL) >> 56 | \
95 ((x) & 0x000000000000ff00ULL) << 40 | \
96 ((x) & 0x00ff000000000000ULL) >> 40 | \
97 ((x) & 0x0000000000ff0000ULL) << 24 | \
98 ((x) & 0x0000ff0000000000ULL) >> 24 | \
99 ((x) & 0x00000000ff000000ULL) << 8 | \
100 ((x) & 0x000000ff00000000ULL) >> 8)
101
e39b673e 102#if !defined(__KLIBC__)
efd441d1
NB
103#if BYTE_ORDER == LITTLE_ENDIAN
104#define __cpu_to_le16(_x) (_x)
105#define __cpu_to_le32(_x) (_x)
106#define __cpu_to_le64(_x) (_x)
107#define __le16_to_cpu(_x) (_x)
108#define __le32_to_cpu(_x) (_x)
109#define __le64_to_cpu(_x) (_x)
974e620d
NB
110
111#define __cpu_to_be16(_x) bswap_16(_x)
112#define __cpu_to_be32(_x) bswap_32(_x)
113#define __cpu_to_be64(_x) bswap_64(_x)
114#define __be16_to_cpu(_x) bswap_16(_x)
115#define __be32_to_cpu(_x) bswap_32(_x)
116#define __be64_to_cpu(_x) bswap_64(_x)
efd441d1
NB
117#elif BYTE_ORDER == BIG_ENDIAN
118#define __cpu_to_le16(_x) bswap_16(_x)
119#define __cpu_to_le32(_x) bswap_32(_x)
120#define __cpu_to_le64(_x) bswap_64(_x)
121#define __le16_to_cpu(_x) bswap_16(_x)
122#define __le32_to_cpu(_x) bswap_32(_x)
123#define __le64_to_cpu(_x) bswap_64(_x)
974e620d
NB
124
125#define __cpu_to_be16(_x) (_x)
126#define __cpu_to_be32(_x) (_x)
127#define __cpu_to_be64(_x) (_x)
128#define __be16_to_cpu(_x) (_x)
129#define __be32_to_cpu(_x) (_x)
130#define __be64_to_cpu(_x) (_x)
1e0d770c
NB
131#else
132# error "unknown endianness."
133#endif
e39b673e 134#endif /* __KLIBC__ */
1e0d770c
NB
135
136
137
4b1ac34b
NB
138/* general information that might be extracted from a superblock */
139struct mdinfo {
140 mdu_array_info_t array;
141 mdu_disk_info_t disk;
142 __u64 events;
fbf8a0b7 143 int uuid[4];
31317663 144 char name[33];
353632d9 145 unsigned long long data_offset;
598f0d58
NB
146 unsigned long long component_size; /* same as array.size, except in
147 * sectors and up to 64bits.
148 */
da9b4a62
DW
149 unsigned long long custom_array_size; /* size for non-default sized
150 * arrays (in sectors)
151 */
353632d9
NB
152 int reshape_active;
153 unsigned long long reshape_progress;
0fd5c350 154 unsigned long long resync_start;
a67dd8cc 155 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 156 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 157 int errors;
7e0f6979
NB
158 int cache_size; /* size of raid456 stripe cache*/
159 int mismatch_cnt;
160 char text_version[50];
dd15dc4a
NB
161
162 int container_member; /* for assembling external-metatdata arrays
163 * This is to be used internally by metadata
164 * handler only */
06c7f68e
NB
165
166 char sys_name[20];
7e0f6979 167 struct mdinfo *devs;
06c7f68e 168 struct mdinfo *next;
549e9569
NB
169
170 /* Device info for mdmon: */
171 int state_fd;
8d45d196
DW
172 #define DS_FAULTY 1
173 #define DS_INSYNC 2
174 #define DS_WRITE_MOSTLY 4
175 #define DS_SPARE 8
176 #define DS_BLOCKED 16
177 #define DS_REMOVE 1024
92967543 178 #define DS_UNBLOCK 2048
549e9569
NB
179 int prev_state, curr_state, next_state;
180
4b1ac34b
NB
181};
182
5bbb4842
NB
183struct createinfo {
184 int uid;
185 int gid;
186 int autof;
187 int mode;
38098016 188 int symlinks;
058574b1 189 struct supertype *supertype;
5bbb4842
NB
190};
191
9a9dab36 192#define Name "mdadm"
682c7051 193
e0d19036
NB
194enum mode {
195 ASSEMBLE=1,
196 BUILD,
197 CREATE,
198 MANAGE,
199 MISC,
200 MONITOR,
dd0781e5 201 GROW,
8382f19b 202 INCREMENTAL,
1f48664b 203 AUTODETECT,
e0d19036
NB
204};
205
64c4757e 206extern char short_options[];
024768c4 207extern char short_bitmap_options[];
c06487ce 208extern char short_bitmap_auto_options[];
64c4757e 209extern struct option long_options[];
56eedc1a 210extern char Version[], Usage[], Help[], OptionHelp[],
dd0781e5 211 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 212 Help_incr[],
e0d19036 213 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 214
997aed5d
NB
215/* for option that don't have short equivilents, we assign arbitrary
216 * small numbers. '1' means an undecorated option, so we start at '2'.
217 */
218enum special_options {
219 AssumeClean = 2,
220 BitmapChunk,
221 WriteBehind,
222 ReAdd,
223 NoDegraded,
224 Sparc22,
225 BackupFile,
226 HomeHost,
589395d6 227 AutoHomeHost,
38098016 228 Symlinks,
1f48664b 229 AutoDetect,
1770662b 230 Waitclean,
4cce4069 231 DetailPlatform,
997aed5d
NB
232};
233
64c4757e 234/* structures read from config file */
52826846
NB
235/* List of mddevice names and identifiers
236 * Identifiers can be:
237 * uuid=128-hex-uuid
238 * super-minor=decimal-minor-number-from-superblock
239 * devices=comma,separated,list,of,device,names,with,wildcards
240 *
241 * If multiple fields are present, the intersection of all matching
242 * devices is considered
243 */
98c6faba 244#define UnSet (0xfffe)
52826846 245typedef struct mddev_ident_s {
dd0781e5 246 char *devname;
aba69144 247
dd0781e5 248 int uuid_set;
3fa06e9d 249 int uuid[4];
947fd4dd 250 char name[33];
52826846 251
98c6faba 252 unsigned int super_minor;
52826846 253
dd0781e5 254 char *devices; /* comma separated list of device
52826846
NB
255 * names with wild cards
256 */
dd0781e5 257 int level;
98c6faba
NB
258 unsigned int raid_disks;
259 unsigned int spare_disks;
82d9eba6 260 struct supertype *st;
dd0781e5
NB
261 int autof; /* 1 for normal, 2 for partitioned */
262 char *spare_group;
7ef02d01 263 char *bitmap_file;
c82f047c 264 int bitmap_fd;
dd0781e5 265
1771a6e2
N
266 char *container; /* /dev/whatever name of container, or
267 * uuid of container. You would expect
268 * this to be the 'devname' or UUID
dbb44303
N
269 * of some other entry.
270 */
271 char *member; /* subarray within a container */
272
52826846
NB
273 struct mddev_ident_s *next;
274} *mddev_ident_t;
64c4757e
NB
275
276/* List of device names - wildcards expanded */
277typedef struct mddev_dev_s {
278 char *devname;
cd29a5c8
NB
279 char disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
280 * Not set for names read from .config
281 */
b3d31955 282 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
fe80f49b 283 char re_add;
811e6cbe 284 char used; /* set when used */
9008ed1c
N
285 struct mdinfo *content; /* If devname is a container, this might list
286 * the remaining member arrays. */
64c4757e
NB
287 struct mddev_dev_s *next;
288} *mddev_dev_t;
289
682c7051
NB
290typedef struct mapping {
291 char *name;
292 int num;
293} mapping_t;
294
e0d19036
NB
295
296struct mdstat_ent {
297 char *dev;
298 int devnum;
299 int active;
300 char *level;
301 char *pattern; /* U or up, _ for down */
302 int percent; /* -1 if no resync */
e5329c37 303 int resync; /* 1 if resync, 0 if recovery */
549e9569
NB
304 int devcnt;
305 int raid_disks;
306 int chunk_size;
307 char * metadata_version;
e0d19036
NB
308 struct mdstat_ent *next;
309};
310
22a88995 311extern struct mdstat_ent *mdstat_read(int hold, int start);
e0d19036 312extern void free_mdstat(struct mdstat_ent *ms);
dd0781e5 313extern void mdstat_wait(int seconds);
1ed3f387 314extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
8382f19b
NB
315extern int mddev_busy(int devnum);
316
317struct map_ent {
318 struct map_ent *next;
319 int devnum;
1522c538 320 char metadata[20];
8382f19b 321 int uuid[4];
195254b8 322 int bad;
8382f19b
NB
323 char *path;
324};
1522c538 325extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
8382f19b
NB
326 int uuid[4], char *path);
327extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
f2e55ecc
N
328extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
329extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
330extern void map_read(struct map_ent **melp);
331extern int map_write(struct map_ent *mel);
332extern void map_delete(struct map_ent **mapp, int devnum);
333extern void map_free(struct map_ent *map);
334extern void map_add(struct map_ent **melp,
1522c538 335 int devnum, char *metadata, int uuid[4], char *path);
ad5bc697
N
336extern int map_lock(struct map_ent **melp);
337extern void map_unlock(struct map_ent **melp);
e0d19036 338
e86c9dd6 339/* various details can be requested */
dab4a513
DW
340enum sysfs_read_flags {
341 GET_LEVEL = (1 << 0),
342 GET_LAYOUT = (1 << 1),
343 GET_COMPONENT = (1 << 2),
344 GET_CHUNK = (1 << 3),
345 GET_CACHE = (1 << 4),
346 GET_MISMATCH = (1 << 5),
347 GET_VERSION = (1 << 6),
348 GET_DISKS = (1 << 7),
349 GET_DEGRADED = (1 << 8),
350 GET_SAFEMODE = (1 << 9),
351 GET_DEVS = (1 << 10), /* gets role, major, minor */
352 GET_OFFSET = (1 << 11),
353 GET_SIZE = (1 << 12),
354 GET_STATE = (1 << 13),
355 GET_ERROR = (1 << 14),
356 SKIP_GONE_DEVS = (1 << 15),
357};
e86c9dd6
NB
358
359/* If fd >= 0, get the array it is open on,
360 * else use devnum. >=0 -> major9. <0.....
361 */
549e9569 362extern int sysfs_open(int devnum, char *devname, char *attr);
f35f2525 363extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
7e0f6979
NB
364extern void sysfs_free(struct mdinfo *sra);
365extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
1770662b
DW
366extern int sysfs_attr_match(const char *attr, const char *str);
367extern int sysfs_match_word(const char *word, char **list);
7e0f6979 368extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 369 char *name, char *val);
7e0f6979 370extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 371 char *name, unsigned long long val);
97590376 372extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
373extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
374 char *name);
375extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 376extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 377 char *name, unsigned long long *val);
7236ee7a 378extern int sysfs_fd_get_str(int fd, char *val, int size);
93ecfa01
N
379extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
380 char *name, char *val, int size);
8ed3e5e1 381extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 382extern int sysfs_set_array(struct mdinfo *info, int vers);
462906cd
N
383extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd,
384 int in_sync);
f1665f72 385extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
f94d52f4 386extern int sysfs_unique_holder(int devnum, long rdev);
c69b251b 387extern int load_sys(char *path, char *buf);
e86c9dd6
NB
388
389
390extern int save_stripes(int *source, unsigned long long *offsets,
391 int raid_disks, int chunk_size, int level, int layout,
392 int nwrites, int *dest,
a6288483
N
393 unsigned long long start, unsigned long long length,
394 char *buf);
353632d9
NB
395extern int restore_stripes(int *dest, unsigned long long *offsets,
396 int raid_disks, int chunk_size, int level, int layout,
397 int source, unsigned long long read_offset,
398 unsigned long long start, unsigned long long length);
e86c9dd6 399
52826846
NB
400#ifndef Sendmail
401#define Sendmail "/usr/lib/sendmail -t"
402#endif
403
773135f5
NB
404#define SYSLOG_FACILITY LOG_DAEMON
405
682c7051
NB
406extern char *map_num(mapping_t *map, int num);
407extern int map_name(mapping_t *map, char *name);
b640a252 408extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
682c7051 409
16c6fa80 410extern char *map_dev(int major, int minor, int create);
64c4757e 411
549e9569 412struct active_array;
2e735d19 413struct metadata_update;
64c4757e 414
6adfd3af
NB
415/* A superswitch provides entry point the a metadata handler.
416 *
417 * The super_switch primarily operates on some "metadata" that
418 * is accessed via the 'supertype'.
419 * This metadata has one of three possible sources.
420 * 1/ It is read from a single device. In this case it may not completely
421 * describe the array or arrays as some information might be on other
422 * devices.
423 * 2/ It is read from all devices in a container. In this case all
424 * information is present.
425 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
426 * be complete once enough ->add_to_super calls have completed.
427 *
428 * When creating an array inside a container, the metadata will be
429 * formed by a combination of 2 and 3. The metadata or the array is read,
430 * then new information is added.
431 *
432 * The metadata must sometimes have a concept of a 'current' array
433 * and a 'current' device.
434 * The 'current' array is set by init_super to be the newly created array,
435 * or is set by super_by_fd when it finds it is looking at an array inside
436 * a container.
437 *
438 * The 'current' device is either the device that the metadata was read from
439 * in case 1, or the last device added by add_to_super in case 3.
440 * Case 2 does not identify a 'current' device.
441 */
f9ce90ba 442extern struct superswitch {
6adfd3af
NB
443
444 /* Used to report details of metadata read from a component
445 * device. ->load_super has been called.
446 */
3da92f27 447 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 448 void (*brief_examine_super)(struct supertype *st, int verbose);
0d726f17 449 void (*export_examine_super)(struct supertype *st);
6adfd3af
NB
450
451 /* Used to report details of an active array.
452 * ->load_super was possibly given a 'component' string.
453 */
3da92f27 454 void (*detail_super)(struct supertype *st, char *homehost);
3da92f27 455 void (*brief_detail_super)(struct supertype *st);
0d726f17 456 void (*export_detail_super)(struct supertype *st);
6adfd3af 457
4cce4069 458 /* Optional: platform hardware / firmware details */
5615172f 459 int (*detail_platform)(int verbose, int enumerate_only);
4cce4069 460
6adfd3af
NB
461 /* Used:
462 * to get uuid to storing in bitmap metadata
463 * and 'reshape' backup-data metadata
464 * To see if a device is being re-added to an array it was part of.
465 */
3da92f27 466 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 467
8592f29d 468 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
469 * the container, or about an individual array within the container.
470 * The determination is made either by:
471 * load_super being given a 'component' string.
472 * validate_geometry determining what to create.
d2ca6449
NB
473 * The info includes both array information and device information.
474 * The particular device should be:
475 * The last device added by add_to_super
476 * The device the metadata was loaded from by load_super
6adfd3af 477 */
3da92f27 478 void (*getinfo_super)(struct supertype *st, struct mdinfo *info);
6adfd3af
NB
479
480 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 481 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 482 */
3da92f27 483 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
484
485 /* Make one of several generic modifications to metadata
486 * prior to assembly (or other times).
487 * sparc2.2 - first bug in early 0.90 metadata
488 * super-minor - change name of 0.90 metadata
489 * summaries - 'correct' any redundant data
490 * resync - mark array as dirty to trigger a resync.
491 * uuid - set new uuid - only 0.90 or 1.x
492 * name - change the name of the array (where supported)
493 * homehost - change which host this array is tied to.
494 * devicesize - If metadata is at start of device, change recorded
495 * device size to match actual device size
496 * byteorder - swap bytes for 0.90 metadata
497 *
498 * force-one - mark that device as uptodate, not old or failed.
499 * force-array - mark array as clean if it would not otherwise
500 * assemble
501 * assemble - not sure how this is different from force-one...
502 * linear-grow-new - add a new device to a linear array, but don't
503 * change the size: so superblock still matches
504 * linear-grow-update - now change the size of the array.
505 */
68c7d6d7 506 int (*update_super)(struct supertype *st, struct mdinfo *info,
3da92f27 507 char *update,
e5eac01f
NB
508 char *devname, int verbose,
509 int uuid_set, char *homehost);
6adfd3af
NB
510
511 /* Create new metadata for new array as described. This could
512 * be a new container, or an array in a pre-existing container.
513 * Also used to zero metadata prior to writing it to invalidate old
514 * metadata.
515 */
3da92f27
NB
516 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
517 unsigned long long size, char *name,
518 char *homehost, int *uuid);
6adfd3af
NB
519
520 /* update the metadata to include new device, either at create or
521 * when hot-adding a spare.
522 */
f20c3968 523 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
111d01fc 524 int fd, char *devname);
6adfd3af
NB
525
526 /* Write metadata to one device when fixing problems or adding
527 * a new device.
528 */
3da92f27 529 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
530
531 /* Write all metadata for this array.
532 */
111d01fc 533 int (*write_init_super)(struct supertype *st);
64557c33 534 int (*compare_super)(struct supertype *st, struct supertype *tst);
3da92f27 535 int (*load_super)(struct supertype *st, int fd, char *devname);
82d9eba6 536 struct supertype * (*match_metadata_desc)(char *arg);
1bf4e2d9 537 __u64 (*avail_size)(struct supertype *st, __u64 size);
3da92f27 538 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 539 int delay, int write_behind,
f9c25f1d 540 unsigned long long size, int may_change, int major);
3da92f27
NB
541 void (*locate_bitmap)(struct supertype *st, int fd);
542 int (*write_bitmap)(struct supertype *st, int fd);
543 void (*free_super)(struct supertype *st);
78e44928
NB
544
545 /* validate_geometry is called with an st returned by
546 * match_metadata_desc.
547 * It should check that the geometry described in compatible with
548 * the metadata type. It will be called repeatedly as devices
549 * added to validate changing size and new devices. If there are
550 * inter-device dependencies, it should record sufficient details
551 * so these can be validated.
e46273eb 552 * Both 'size' and '*freesize' are in sectors. chunk is bytes.
78e44928 553 */
17f25ca6
NB
554 int (*validate_geometry)(struct supertype *st, int level, int layout,
555 int raiddisks,
556 int chunk, unsigned long long size,
2c514b71
NB
557 char *subdev, unsigned long long *freesize,
558 int verbose);
598f0d58
NB
559
560 struct mdinfo *(*container_content)(struct supertype *st);
a18a888e
DW
561 /* Allow a metadata handler to override mdadm's default layouts */
562 int (*default_layout)(int level); /* optional */
598f0d58 563
549e9569 564/* for mdmon */
cba0191b
NB
565 int (*open_new)(struct supertype *c, struct active_array *a,
566 char *inst);
ed9d66aa
NB
567
568 /* Tell the metadata handler the current state of the array.
569 * This covers whether it is known to be consistent (no pending writes)
01f157d7 570 * and how far along a resync is known to have progressed
ed9d66aa
NB
571 * (in a->resync_start).
572 * resync status is really irrelevant if the array is not consistent,
573 * but some metadata (DDF!) have a place to record the distinction.
01f157d7
N
574 * If 'consistent' is '2', then the array can mark it dirty if a
575 * resync/recovery/whatever is required, or leave it clean if not.
576 * Return value is 0 dirty (not consistent) and 1 if clean.
577 * it is only really important if consistent is passed in as '2'.
4e5528c6 578 */
01f157d7 579 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
580
581 /* When the state of a device might have changed, we call set_disk to
582 * tell the metadata what the current state is.
583 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
584 * transitions.
585 * set_disk might be called when the state of the particular disk has
586 * not in fact changed.
587 */
8d45d196 588 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
589 void (*sync_metadata)(struct supertype *st);
590 void (*process_update)(struct supertype *st,
591 struct metadata_update *update);
edd8d13c
NB
592 void (*prepare_update)(struct supertype *st,
593 struct metadata_update *update);
549e9569 594
6c3fb95c
NB
595 /* activate_spare will check if the array is degraded and, if it
596 * is, try to find some spare space in the container.
597 * On success, it add appropriate updates (For process_update) to
598 * to the 'updates' list and returns a list of 'mdinfo' identifying
599 * the device, or devices as there might be multiple missing
600 * devices and multiple spares available.
601 */
602 struct mdinfo *(*activate_spare)(struct active_array *a,
603 struct metadata_update **updates);
549e9569 604
f277ce36 605 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 606 int external;
4cce4069 607 const char *name; /* canonical metadata name */
0063ecba 608} super0, super1, super_ddf, *superlist[];
82d9eba6 609
0063ecba 610extern struct superswitch super_imsm;
cdddbdbc 611
edd8d13c
NB
612struct metadata_update {
613 int len;
614 char *buf;
615 void *space; /* allocated space that monitor will use */
616 struct metadata_update *next;
617};
618
6adfd3af
NB
619/* A supertype holds a particular collection of metadata.
620 * It identifies the metadata type by the superswitch, and the particular
621 * sub-version of that metadata type.
622 * metadata read in or created is stored in 'sb' and 'info'.
623 * There are also fields used by mdmon to track containers.
624 *
352452c3
N
625 * A supertype may refer to:
626 * Just an array, possibly in a container
627 * A container, not identifying any particular array
628 * Info read from just one device, not yet fully describing the array/container.
629 *
630 *
6adfd3af
NB
631 * A supertype is created by:
632 * super_by_fd
633 * guess_super
634 * dup_super
635 */
82d9eba6
NB
636struct supertype {
637 struct superswitch *ss;
638 int minor_version;
ea329559 639 int max_devs;
2f6079dc 640 int container_dev; /* devnum of container */
f7e7067b 641 char subarray[32]; /* name of array inside container */
64557c33 642 void *sb;
111d01fc 643 void *info;
352452c3
N
644 int loaded_container; /* Set if load_super found a container,
645 * not just one device */
549e9569 646
edd8d13c
NB
647 struct metadata_update *updates;
648 struct metadata_update **update_tail;
649
549e9569
NB
650 /* extra stuff used by mdmon */
651 struct active_array *arrays;
549e9569 652 int sock; /* listen to external programs */
549e9569
NB
653 int devnum;
654 char *devname; /* e.g. md0. This appears in metadata_verison:
655 * external:/md0/12
656 */
657 int devcnt;
658
659 struct mdinfo *devs;
660
82d9eba6 661};
f9ce90ba 662
1686dc25 663extern struct supertype *super_by_fd(int fd);
82d9eba6 664extern struct supertype *guess_super(int fd);
3da92f27 665extern struct supertype *dup_super(struct supertype *st);
beae1dfe 666extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
8382f19b
NB
667extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
668 mdu_disk_info_t *disk);
a7c6e3fb 669void wait_for(char *dev, int fd);
f9ce90ba 670
838acbc2
NB
671#if __GNUC__ < 3
672struct stat64;
673#endif
674
45e878bb
NB
675#define HAVE_NFTW we assume
676#define HAVE_FTW
677
2df1f269 678#ifdef __UCLIBC__
45e878bb 679# include <features.h>
2df1f269
BRF
680# ifndef __UCLIBC_HAS_LFS__
681# define lseek64 lseek
682# endif
45e878bb
NB
683# ifndef __UCLIBC_HAS_FTW__
684# undef HAVE_FTW
685# undef HAVE_NFTW
686# endif
687#endif
688
689#ifdef __dietlibc__
690# undef HAVE_NFTW
691#endif
692
e39b673e 693#if defined(__KLIBC__)
694# undef HAVE_NFTW
695# undef HAVE_FTW
696#endif
697
45e878bb 698#ifndef HAVE_NFTW
173fc515 699# define FTW_PHYS 1
45e878bb
NB
700# ifndef HAVE_FTW
701 struct FTW {};
173fc515
NB
702# endif
703#endif
704
45e878bb
NB
705#ifdef HAVE_FTW
706# include <ftw.h>
707#endif
708
173fc515
NB
709extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
710
f9ce90ba 711
64c4757e 712extern int Manage_ro(char *devname, int fd, int readonly);
91f068bf 713extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
dd0781e5 714extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
64c4757e 715extern int Manage_subdevs(char *devname, int fd,
dab6685f 716 mddev_dev_t devlist, int verbose);
1f48664b 717extern int autodetect(void);
e5329c37 718extern int Grow_Add_device(char *devname, int fd, char *newdev);
8fac0577 719extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
06b0d786 720extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
e86c9dd6 721 long long size,
19678e53 722 int level, char *layout_str, int chunksize, int raid_disks);
06b0d786
NB
723extern int Grow_restart(struct supertype *st, struct mdinfo *info,
724 int *fdlist, int cnt, char *backup_file);
64c4757e
NB
725
726
7f91af49 727extern int Assemble(struct supertype *st, char *mddev,
52826846 728 mddev_ident_t ident,
06b0d786 729 mddev_dev_t devlist, char *backup_file,
64c4757e 730 int readonly, int runstop,
0ac91628 731 char *update, char *homehost, int require_homehost,
64c4757e
NB
732 int verbose, int force);
733
7f91af49
N
734extern int Build(char *mddev, int chunk, int level, int layout,
735 int raiddisks, mddev_dev_t devlist, int assume_clean,
736 char *bitmap_file, int bitmap_chunk, int write_behind,
83208785 737 int delay, int verbose, int autof, unsigned long long size);
64c4757e
NB
738
739
7f91af49 740extern int Create(struct supertype *st, char *mddev,
5dd497ee 741 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
3d3dd91e 742 char *name, char *homehost, int *uuid,
cd29a5c8 743 int subdevs, mddev_dev_t devlist,
47d79ef8 744 int runstop, int verbose, int force, int assume_clean,
7f91af49 745 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
64c4757e 746
54bad364 747extern int Detail(char *dev, int brief, int export, int test, char *homehost);
4cce4069 748extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
e0d19036 749extern int Query(char *dev);
0d726f17
KS
750extern int Examine(mddev_dev_t devlist, int brief, int export, int scan,
751 int SparcAdjust, struct supertype *forcest, char *homehost);
cd29a5c8 752extern int Monitor(mddev_dev_t devlist,
e0d19036 753 char *mailaddr, char *alert_cmd,
aa88f531 754 int period, int daemonise, int scan, int oneshot,
8aec876d 755 int dosyslog, int test, char *pidfile);
64c4757e 756
111d01fc 757extern int Kill(char *dev, int force, int quiet, int noexcl);
b90c0e9a 758extern int Wait(char *dev);
27dec8fa 759extern int WaitClean(char *dev, int verbose);
9a9dab36 760
8382f19b 761extern int Incremental(char *devname, int verbose, int runstop,
0ac91628
N
762 struct supertype *st, char *homehost, int require_homehost,
763 int autof);
598f0d58 764extern int Incremental_container(struct supertype *st, char *devname,
215bb3f7
N
765 int verbose, int runstop, int autof,
766 int trustworthy);
8382f19b
NB
767extern void RebuildMap(void);
768extern int IncrementalScan(int verbose);
769
c82f047c
NB
770extern int CreateBitmap(char *filename, int force, char uuid[16],
771 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 772 unsigned long write_behind,
dcec9ee5
NB
773 unsigned long long array_size,
774 int major);
55935d51 775extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
bf4fb153 776extern int bitmap_update_uuid(int fd, int *uuid, int swap);
2a528478 777extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
c82f047c 778
64c4757e 779extern int md_get_version(int fd);
e0d19036 780extern int get_linux_version(void);
84e11361 781extern long long parse_size(char *size);
64c4757e 782extern int parse_uuid(char *str, int uuid[4]);
4a06e2c2
N
783extern int parse_layout_10(char *layout);
784extern int parse_layout_faulty(char *layout);
682c7051
NB
785extern int check_ext2(int fd, char *name);
786extern int check_reiser(int fd, char *name);
787extern int check_raid(int fd, char *name);
64c4757e 788
dd0781e5 789extern int get_mdp_major(void);
8b0dabea 790extern int dev_open(char *dev, int flags);
e8a70c89 791extern int open_dev(int devnum);
a931db9e 792extern int open_dev_excl(int devnum);
8d80900b 793extern int is_standard(char *dev, int *nump);
9008ed1c 794extern int same_dev(char *one, char *two);
dd0781e5 795
f1ae21c4 796extern int parse_auto(char *str, char *msg, int config);
8aec876d
NB
797extern mddev_ident_t conf_get_ident(char *dev);
798extern mddev_dev_t conf_get_devs(void);
8382f19b 799extern int conf_test_dev(char *devname);
31015d57 800extern int conf_test_metadata(const char *version);
8aec876d
NB
801extern struct createinfo *conf_get_create_info(void);
802extern void set_conffile(char *file);
803extern char *conf_get_mailaddr(void);
804extern char *conf_get_mailfrom(void);
805extern char *conf_get_program(void);
0ac91628 806extern char *conf_get_homehost(int *require_homehostp);
e0d19036 807extern char *conf_line(FILE *file);
98c6faba 808extern char *conf_word(FILE *file, int allow_key);
0ac91628
N
809extern int conf_name_is_free(char *name);
810extern int devname_matches(char *name, char *match);
360b4636 811extern struct mddev_ident_s *conf_match(struct mdinfo *info, struct supertype *st);
0ac91628 812
e0d19036
NB
813extern void free_line(char *line);
814extern int match_oneof(char *devices, char *devname);
e0d19036 815extern void uuid_from_super(int uuid[4], mdp_super_t *super);
36ba7d48 816extern const int uuid_match_any[4];
f277ce36 817extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 818extern void copy_uuid(void *a, int b[4], int swapuuid);
d7288ddc 819extern char *fname_from_uuid(struct supertype *st,
ff54de6e 820 struct mdinfo *info, char *buf, char sep);
4b1ac34b 821extern unsigned long calc_csum(void *super, int bytes);
583315d9 822extern int enough(int level, int raid_disks, int layout, int clean,
265e0f17 823 char *avail, int avail_disks);
e0d19036 824extern int ask(char *mesg);
8fac0577 825extern unsigned long long get_component_size(int fd);
0430ed48 826extern void remove_partitions(int fd);
5f8097be
NB
827extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
828 int chunksize, unsigned long long devsize);
edd8d13c
NB
829extern int flush_metadata_updates(struct supertype *st);
830extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c
DW
831extern int assemble_container_content(struct supertype *st, int mdfd,
832 struct mdinfo *content, int runstop,
833 char *chosen_name, int verbose);
e0d19036 834
7801ac20
N
835extern int add_disk(int mdfd, struct supertype *st,
836 struct mdinfo *sra, struct mdinfo *info);
f35f2525 837extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
cd29a5c8 838
56eb10c0 839extern char *human_size(long long bytes);
e4965ef8
N
840extern char *human_size_brief(long long bytes);
841extern void print_r10_layout(int layout);
e0d19036 842
63152c1b
NB
843#define NoMdDev (1<<23)
844extern int find_free_devnum(int use_partitions);
845
e0d19036
NB
846extern void put_md_name(char *name);
847extern char *get_md_name(int dev);
848
c913b90e 849extern char DefaultConfFile[];
b5e64645 850
69207ff6
N
851extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
852 char *chosen);
853/* values for 'trustworthy' */
854#define LOCAL 1
855#define FOREIGN 2
856#define METADATA 3
6be1d39d 857extern int open_mddev(char *dev, int report_errors);
a322f70c 858extern int open_container(int fd);
b5e64645 859
a931db9e
NB
860extern int mdmon_running(int devnum);
861extern int signal_mdmon(int devnum);
40ebbb9c 862extern int check_env(char *name);
8850ee3e 863extern int start_mdmon(int devnum);
a931db9e 864
2f6079dc 865extern char *devnum2devname(int num);
77472ff8 866extern int devname2devnum(char *name);
c94709e8 867extern int stat2devnum(struct stat *st);
2f6079dc 868extern int fd2devnum(int fd);
b5e64645 869
a931db9e
NB
870static inline int dev2major(int d)
871{
872 if (d >= 0)
873 return MD_MAJOR;
874 else
875 return get_mdp_major();
876}
877
878static inline int dev2minor(int d)
879{
880 if (d >= 0)
881 return d;
882 return (-1-d) << MdpMinorShift;
883}
884
6416d527
NB
885static inline int ROUND_UP(int a, int base)
886{
887 return ((a+base-1)/base)*base;
888}
889
3c558363
N
890static inline int is_subarray(char *vers)
891{
e9dd1598
N
892 /* The version string for a 'subarray' (an array in a container)
893 * is
894 * /containername/componentname for normal read-write arrays
895 * -containername/componentname for read-only arrays.
896 * containername is e.g. md0, md_d1
897 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
898 */
899 return (*vers == '/' || *vers == '-');
3c558363
N
900}
901
2a24d7b6
DW
902#ifdef DEBUG
903#define dprintf(fmt, arg...) \
904 fprintf(stderr, fmt, ##arg)
905#else
906#define dprintf(fmt, arg...) \
907 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
908#endif
1a0ee0ba
DK
909#include <assert.h>
910#include <stdarg.h>
911static inline int xasprintf(char **strp, const char *fmt, ...) {
912 va_list ap;
913 int ret;
914 va_start(ap, fmt);
089485cb 915 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
916 va_end(ap);
917 assert(ret >= 0);
918 return ret;
919}
2a24d7b6 920
b5e64645
NB
921#define LEVEL_MULTIPATH (-4)
922#define LEVEL_LINEAR (-1)
923#define LEVEL_FAULTY (-5)
924
17f25ca6
NB
925/* kernel module doesn't know about these */
926#define LEVEL_CONTAINER (-100)
a322f70c 927#define LEVEL_UNSUPPORTED (-200)
17f25ca6 928
b5e64645
NB
929
930/* faulty stuff */
931
932#define WriteTransient 0
933#define ReadTransient 1
934#define WritePersistent 2
935#define ReadPersistent 3
936#define WriteAll 4 /* doesn't go to device */
937#define ReadFixable 5
938#define Modes 6
939
940#define ClearErrors 31
941#define ClearFaults 30
942
943#define AllPersist 100 /* internal use only */
944#define NoPersist 101
945
946#define ModeMask 0x1f
947#define ModeShift 5
fe6729fa
NB
948
949
950#ifdef __TINYC__
951#undef minor
952#undef major
953#undef makedev
954#define minor(x) ((x)&0xff)
955#define major(x) (((x)>>8)&0xff)
956#define makedev(M,m) (((M)<<8) | (m))
957#endif
958
b640a252 959/* for raid4/5/6 */
e86c9dd6
NB
960#define ALGORITHM_LEFT_ASYMMETRIC 0
961#define ALGORITHM_RIGHT_ASYMMETRIC 1
962#define ALGORITHM_LEFT_SYMMETRIC 2
963#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
964
965/* Define non-rotating (raid4) algorithms. These allow
966 * conversion of raid4 to raid5.
967 */
968#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
969#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
970
971/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
972 * Firstly, the exact positioning of the parity block is slightly
973 * different between the 'LEFT_*' modes of md and the "_N_*" modes
974 * of DDF.
975 * Secondly, or order of datablocks over which the Q syndrome is computed
976 * is different.
977 * Consequently we have different layouts for DDF/raid6 than md/raid6.
978 * These layouts are from the DDFv1.2 spec.
979 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
980 * leaves RLQ=3 as 'Vendor Specific'
981 */
982
983#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
984#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
985#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
986
987
988/* For every RAID5 algorithm we define a RAID6 algorithm
989 * with exactly the same layout for data and parity, and
990 * with the Q block always on the last device (N-1).
991 * This allows trivial conversion from RAID5 to RAID6
992 */
993#define ALGORITHM_LEFT_ASYMMETRIC_6 16
994#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
995#define ALGORITHM_LEFT_SYMMETRIC_6 18
996#define ALGORITHM_RIGHT_SYMMETRIC_6 19
997#define ALGORITHM_PARITY_0_6 20
998#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
999