]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
Manage: check device policies in manage_add_external()
[thirdparty/mdadm.git] / mdadm.h
CommitLineData
64c4757e 1/*
9a9dab36 2 * mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e 3 *
e736b623 4 * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
64c4757e
NB
5 *
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 * Author: Neil Brown
e736b623 22 * Email: <neilb@suse.de>
64c4757e
NB
23 */
24
4ccdb956 25#define _GNU_SOURCE
c2c9bb6f 26#define _FILE_OFFSET_BITS 64
64c4757e 27#include <unistd.h>
0d35d5c4 28#ifdef __GLIBC__
64c4757e 29extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
0d35d5c4 30#elif !defined(lseek64)
fffdbe5e 31# if defined(__NO_STAT64) || __WORDSIZE != 32
f783ca4f
NB
32# define lseek64 lseek
33# endif
98c6faba 34#endif
64c4757e 35
7fcbfd7c 36#include <assert.h>
64c4757e
NB
37#include <sys/types.h>
38#include <sys/stat.h>
7fcbfd7c 39#include <stdarg.h>
d15a1f72 40#include <stdint.h>
64c4757e
NB
41#include <stdlib.h>
42#include <time.h>
11018a45 43#include <sys/time.h>
64c4757e
NB
44#include <getopt.h>
45#include <fcntl.h>
46#include <stdio.h>
47#include <errno.h>
48#include <string.h>
773135f5 49#include <syslog.h>
a35aa68f 50#include <stdbool.h>
83a379cf 51#include <signal.h>
977d12d7
JS
52/* Newer glibc requires sys/sysmacros.h directly for makedev() */
53#include <sys/sysmacros.h>
b56c3630
NB
54#ifdef __dietlibc__
55#include <strings.h>
280a927d
NB
56/* dietlibc has deprecated random and srandom!! */
57#define random rand
58#define srandom srand
98c6faba
NB
59#endif
60
e80357f8
GJ
61#ifdef NO_COROSYNC
62#define CS_OK 1
63typedef uint64_t cmap_handle_t;
64#else
65#include <corosync/cmap.h>
66#endif
67
d15a1f72
GJ
68#ifndef NO_DLM
69#include <libdlm.h>
70#include <errno.h>
71#else
72#define LKF_NOQUEUE 0x00000001
d15a1f72
GJ
73#define LKM_PWMODE 4
74#define EUNLOCK 0x10002
75
76typedef void *dlm_lshandle_t;
77
78struct dlm_lksb {
79 int sb_status;
80 uint32_t sb_lkid;
81 char sb_flags;
82 char *sb_lvbptr;
83};
84#endif
85
64c4757e 86#include <linux/kdev_t.h>
e0d19036
NB
87/*#include <linux/fs.h> */
88#include <sys/mount.h>
89#include <asm/types.h>
64c4757e
NB
90#include <sys/ioctl.h>
91#define MD_MAJOR 9
dd0781e5 92#define MdpMinorShift 6
64c4757e 93
e0d19036 94#ifndef BLKGETSIZE64
98c6faba 95#define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
e0d19036 96#endif
56eb10c0 97
c21e737b 98#define DEFAULT_CHUNK 512
c82f047c
NB
99#define DEFAULT_BITMAP_CHUNK 4096
100#define DEFAULT_BITMAP_DELAY 5
dfd4d8ee 101#define DEFAULT_MAX_WRITE_BEHIND 256
64c4757e 102
0f8347d4
MT
103/* DEV_NUM_PREF is a subpath to numbered MD devices, e.g. /dev/md1 or directory name.
104 * DEV_NUM_PREF_LEN is a length with Null byte excluded.
105 */
106#ifndef DEV_NUM_PREF
107#define DEV_NUM_PREF "/dev/md"
108#define DEV_NUM_PREF_LEN (sizeof(DEV_NUM_PREF) - 1)
109#endif /* DEV_NUM_PREF */
110
b9ce7ab0
MT
111/* DEV_MD_DIR points to named MD devices directory.
112 * DEV_MD_DIR_LEN is a length with Null byte excluded.
113 */
114#ifndef DEV_MD_DIR
115#define DEV_MD_DIR "/dev/md/"
116#define DEV_MD_DIR_LEN (sizeof(DEV_MD_DIR) - 1)
117#endif /* DEV_MD_DIR */
118
753cf905 119/* MAP_DIR should be somewhere that persists across the pivotroot
5d4d1b26 120 * from early boot to late boot.
96fd06ed 121 * /run seems to have emerged as the best standard.
5d4d1b26 122 */
753cf905 123#ifndef MAP_DIR
96fd06ed 124#define MAP_DIR "/run/mdadm"
753cf905
DL
125#endif /* MAP_DIR */
126/* MAP_FILE is what we name the map file we put in MAP_DIR, in case you
127 * want something other than the default of "map"
128 */
129#ifndef MAP_FILE
130#define MAP_FILE "map"
131#endif /* MAP_FILE */
132/* MDMON_DIR is where pid and socket files used for communicating
96fd06ed 133 * with mdmon normally live. Best is /var/run/mdadm as
753cf905
DL
134 * mdmon is needed at early boot then it needs to write there prior
135 * to /var/run being mounted read/write, and it also then needs to
136 * persist beyond when /var/run is mounter read-only. So, to be
137 * safe, the default is somewhere that is read/write early in the
138 * boot process and stays up as long as possible during shutdown.
139 */
140#ifndef MDMON_DIR
96fd06ed 141#define MDMON_DIR "/run/mdadm"
753cf905 142#endif /* MDMON_DIR */
5d4d1b26 143
403410eb
PC
144/* FAILED_SLOTS is where to save files storing recent removal of array
145 * member in order to allow future reuse of disk inserted in the same
146 * slot for array recovery
147 */
148#ifndef FAILED_SLOTS_DIR
96fd06ed 149#define FAILED_SLOTS_DIR "/run/mdadm/failed-slots"
403410eb
PC
150#endif /* FAILED_SLOTS */
151
ff6bb131
MT
152#ifndef MDMON_SERVICE
153#define MDMON_SERVICE "mdmon"
154#endif /* MDMON_SERVICE */
155
156#ifndef GROW_SERVICE
157#define GROW_SERVICE "mdadm-grow-continue"
158#endif /* GROW_SERVICE */
159
64c4757e 160#include "md_u.h"
e0d19036 161#include "md_p.h"
c82f047c 162#include "bitmap.h"
f7dd881f 163#include "msg.h"
64c4757e 164
1e0d770c 165#include <endian.h>
efd441d1
NB
166/* Redhat don't like to #include <asm/byteorder.h>, and
167 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
168 * and there is no standard conversion function so... */
0ae03b8a
NB
169/* And dietlibc doesn't think byteswap is ok, so.. */
170/* #include <byteswap.h> */
dd47b4e0
JS
171#define __mdadm_bswap_16(x) (((x) & 0x00ffU) << 8 | \
172 ((x) & 0xff00U) >> 8)
173#define __mdadm_bswap_32(x) (((x) & 0x000000ffU) << 24 | \
174 ((x) & 0xff000000U) >> 24 | \
175 ((x) & 0x0000ff00U) << 8 | \
176 ((x) & 0x00ff0000U) >> 8)
177#define __mdadm_bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
178 ((x) & 0xff00000000000000ULL) >> 56 | \
179 ((x) & 0x000000000000ff00ULL) << 40 | \
180 ((x) & 0x00ff000000000000ULL) >> 40 | \
181 ((x) & 0x0000000000ff0000ULL) << 24 | \
182 ((x) & 0x0000ff0000000000ULL) >> 24 | \
183 ((x) & 0x00000000ff000000ULL) << 8 | \
184 ((x) & 0x000000ff00000000ULL) >> 8)
0ae03b8a 185
e39b673e 186#if !defined(__KLIBC__)
efd441d1 187#if BYTE_ORDER == LITTLE_ENDIAN
8a5f247d
JS
188#define __cpu_to_le16(_x) (unsigned int)(_x)
189#define __cpu_to_le32(_x) (unsigned int)(_x)
190#define __cpu_to_le64(_x) (unsigned long long)(_x)
191#define __le16_to_cpu(_x) (unsigned int)(_x)
192#define __le32_to_cpu(_x) (unsigned int)(_x)
193#define __le64_to_cpu(_x) (unsigned long long)(_x)
974e620d 194
dd47b4e0
JS
195#define __cpu_to_be16(_x) __mdadm_bswap_16(_x)
196#define __cpu_to_be32(_x) __mdadm_bswap_32(_x)
197#define __cpu_to_be64(_x) __mdadm_bswap_64(_x)
198#define __be16_to_cpu(_x) __mdadm_bswap_16(_x)
199#define __be32_to_cpu(_x) __mdadm_bswap_32(_x)
200#define __be64_to_cpu(_x) __mdadm_bswap_64(_x)
efd441d1 201#elif BYTE_ORDER == BIG_ENDIAN
dd47b4e0
JS
202#define __cpu_to_le16(_x) __mdadm_bswap_16(_x)
203#define __cpu_to_le32(_x) __mdadm_bswap_32(_x)
204#define __cpu_to_le64(_x) __mdadm_bswap_64(_x)
205#define __le16_to_cpu(_x) __mdadm_bswap_16(_x)
206#define __le32_to_cpu(_x) __mdadm_bswap_32(_x)
207#define __le64_to_cpu(_x) __mdadm_bswap_64(_x)
974e620d 208
8a5f247d
JS
209#define __cpu_to_be16(_x) (unsigned int)(_x)
210#define __cpu_to_be32(_x) (unsigned int)(_x)
211#define __cpu_to_be64(_x) (unsigned long long)(_x)
212#define __be16_to_cpu(_x) (unsigned int)(_x)
213#define __be32_to_cpu(_x) (unsigned int)(_x)
214#define __be64_to_cpu(_x) (unsigned long long)(_x)
1e0d770c
NB
215#else
216# error "unknown endianness."
217#endif
e39b673e 218#endif /* __KLIBC__ */
1e0d770c 219
7039d1f8
JS
220/*
221 * Partially stolen from include/linux/unaligned/packed_struct.h
222 */
223struct __una_u16 { __u16 x; } __attribute__ ((packed));
224struct __una_u32 { __u32 x; } __attribute__ ((packed));
225
226static inline __u16 __get_unaligned16(const void *p)
227{
228 const struct __una_u16 *ptr = (const struct __una_u16 *)p;
229 return ptr->x;
230}
231
232static inline __u32 __get_unaligned32(const void *p)
233{
234 const struct __una_u32 *ptr = (const struct __una_u32 *)p;
235 return ptr->x;
236}
237
238static inline void __put_unaligned16(__u16 val, void *p)
239{
240 struct __una_u16 *ptr = (struct __una_u16 *)p;
241 ptr->x = val;
242}
243
244static inline void __put_unaligned32(__u32 val, void *p)
245{
246 struct __una_u32 *ptr = (struct __una_u32 *)p;
247 ptr->x = val;
248}
249
26714713
DD
250/*
251 * Check at compile time that something is of a particular type.
252 * Always evaluates to 1 so you may use it easily in comparisons.
253*/
254
255#define typecheck(type,x) \
256({ type __dummy; \
257 typeof(x) __dummy2; \
258 (void)(&__dummy == &__dummy2); \
259 1; \
260})
261
262/*
263 * These inlines deal with timer wrapping correctly.
264 *
265 * time_after(a,b) returns true if the time a is after time b.
266*/
267
268#define time_after(a,b) \
269 (typecheck(unsigned int, a) && \
270 typecheck(unsigned int, b) && \
271 ((int)((b) - (a)) < 0))
272
273#define time_before(a,b) time_after(b,a)
274
1e5c6983
DW
275/*
276 * min()/max()/clamp() macros that also do
277 * strict type-checking.. See the
278 * "unnecessary" pointer comparison.
279 */
280#define min(x, y) ({ \
281 typeof(x) _min1 = (x); \
282 typeof(y) _min2 = (y); \
283 (void) (&_min1 == &_min2); \
284 _min1 < _min2 ? _min1 : _min2; })
285
286#define max(x, y) ({ \
287 typeof(x) _max1 = (x); \
288 typeof(y) _max2 = (y); \
289 (void) (&_max1 == &_max2); \
290 _max1 > _max2 ? _max1 : _max2; })
1e0d770c 291
1c0aebc2
N
292#define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))
293
577fd104
LG
294#define KIB_TO_BYTES(x) ((x) << 10)
295#define SEC_TO_BYTES(x) ((x) << 9)
296
67417d92
MT
297/**
298 * This is true for native and DDF, IMSM allows 16.
299 */
300#define MD_NAME_MAX 32
301
d56dd607
PB
302extern const char Name[];
303
bb758cca
TM
304struct md_bb_entry {
305 unsigned long long sector;
306 int length;
307};
308
309struct md_bb {
310 int supported;
311 int count;
312 struct md_bb_entry *entries;
313};
314
4b1ac34b
NB
315/* general information that might be extracted from a superblock */
316struct mdinfo {
317 mdu_array_info_t array;
318 mdu_disk_info_t disk;
319 __u64 events;
fbf8a0b7 320 int uuid[4];
31317663 321 char name[33];
353632d9 322 unsigned long long data_offset;
fe384ca0 323 unsigned long long new_data_offset;
598f0d58
NB
324 unsigned long long component_size; /* same as array.size, except in
325 * sectors and up to 64bits.
326 */
da9b4a62
DW
327 unsigned long long custom_array_size; /* size for non-default sized
328 * arrays (in sectors)
329 */
81219e70
LM
330#define NO_RESHAPE 0
331#define VOLUME_RESHAPE 1
332#define CONTAINER_RESHAPE 2
5e88ab2e 333#define RESHAPE_NO_BACKUP 16 /* Mask 'or'ed in */
353632d9
NB
334 int reshape_active;
335 unsigned long long reshape_progress;
6e75048b
AK
336 int recovery_blocked; /* for external metadata it
337 * indicates that there is
338 * reshape in progress in
339 * container,
340 * for native metadata it is
341 * reshape_active field mirror
342 */
051f3265
SL
343 int journal_device_required;
344 int journal_clean;
345
5308f117
AP
346 enum {
347 CONSISTENCY_POLICY_UNKNOWN,
348 CONSISTENCY_POLICY_NONE,
349 CONSISTENCY_POLICY_RESYNC,
350 CONSISTENCY_POLICY_BITMAP,
351 CONSISTENCY_POLICY_JOURNAL,
352 CONSISTENCY_POLICY_PPL,
353 } consistency_policy;
354
80bf9135
N
355 /* During reshape we can sometimes change the data_offset to avoid
356 * over-writing still-valid data. We need to know if there is space.
357 * So getinfo_super will fill in space_before and space_after in sectors.
358 * data_offset can be increased or decreased by this amount.
359 */
360 unsigned long long space_before, space_after;
e1516be1
DW
361 union {
362 unsigned long long resync_start; /* per-array resync position */
363 unsigned long long recovery_start; /* per-device rebuild position */
b7528a20 364 #define MaxSector (~0ULL) /* resync/recovery complete position */
e1516be1 365 };
fbdef498 366 long bitmap_offset; /* 0 == none, 1 == a file */
2432ce9b 367 unsigned int ppl_size;
e97a7cd0 368 int ppl_offset;
a86b1c8d 369 unsigned long long ppl_sector;
a67dd8cc 370 unsigned long safe_mode_delay; /* ms delay to mark clean */
353632d9 371 int new_level, delta_disks, new_layout, new_chunk;
06c7f68e 372 int errors;
f21e18ca 373 unsigned long cache_size; /* size of raid456 stripe cache*/
7e0f6979
NB
374 int mismatch_cnt;
375 char text_version[50];
dd15dc4a
NB
376
377 int container_member; /* for assembling external-metatdata arrays
378 * This is to be used internally by metadata
379 * handler only */
476b00bd
MT
380 /**
381 * flag external handlers can set to indicate that subarrays have:
382 * - not enough disks to start (-1),
383 * - enough disks to start (0),
384 * - all expected disks (1).
385 */
386 int container_enough;
6e6e9874 387 char sys_name[32];
7e0f6979 388 struct mdinfo *devs;
06c7f68e 389 struct mdinfo *next;
549e9569
NB
390
391 /* Device info for mdmon: */
e1516be1 392 int recovery_fd;
549e9569 393 int state_fd;
6dc1785f
TM
394 int bb_fd;
395 int ubb_fd;
8d45d196
DW
396 #define DS_FAULTY 1
397 #define DS_INSYNC 2
398 #define DS_WRITE_MOSTLY 4
399 #define DS_SPARE 8
400 #define DS_BLOCKED 16
401 #define DS_REMOVE 1024
92967543 402 #define DS_UNBLOCK 2048
549e9569 403 int prev_state, curr_state, next_state;
5aa644c6
SL
404
405 /* info read from sysfs */
5e4ca8bb
JS
406 enum {
407 ARRAY_CLEAR,
408 ARRAY_INACTIVE,
409 ARRAY_SUSPENDED,
410 ARRAY_READONLY,
411 ARRAY_READ_AUTO,
412 ARRAY_CLEAN,
413 ARRAY_ACTIVE,
414 ARRAY_WRITE_PENDING,
415 ARRAY_ACTIVE_IDLE,
43ebc910 416 ARRAY_BROKEN,
5e4ca8bb
JS
417 ARRAY_UNKNOWN_STATE,
418 } array_state;
bb758cca 419 struct md_bb bb;
4b1ac34b
NB
420};
421
5bbb4842
NB
422struct createinfo {
423 int uid;
424 int gid;
425 int autof;
426 int mode;
eca944fa 427 int names;
e2efe9e7 428 int bblist;
058574b1 429 struct supertype *supertype;
5bbb4842
NB
430};
431
fbfdcb06 432struct spare_criteria {
0c0f09cb 433 bool criteria_set;
fbfdcb06 434 unsigned long long min_size;
4b57ecf6 435 unsigned int sector_size;
fbfdcb06
AO
436};
437
67417d92
MT
438typedef enum mdadm_status {
439 MDADM_STATUS_SUCCESS = 0,
440 MDADM_STATUS_ERROR,
441 MDADM_STATUS_UNDEF,
442} mdadm_status_t;
443
e0d19036
NB
444enum mode {
445 ASSEMBLE=1,
446 BUILD,
447 CREATE,
448 MANAGE,
449 MISC,
450 MONITOR,
dd0781e5 451 GROW,
8382f19b 452 INCREMENTAL,
1f48664b 453 AUTODETECT,
5187a385 454 mode_count
e0d19036
NB
455};
456
64c4757e 457extern char short_options[];
6c9d9260 458extern char short_monitor_options[];
024768c4 459extern char short_bitmap_options[];
c06487ce 460extern char short_bitmap_auto_options[];
64c4757e 461extern struct option long_options[];
56eedc1a 462extern char Version[], Usage[], Help[], OptionHelp[],
5187a385 463 *mode_help[],
dd0781e5 464 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 465 Help_incr[],
e0d19036 466 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 467
997aed5d 468/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 469 * numbers later than any 'short' character option.
997aed5d
NB
470 */
471enum special_options {
1c7a808c 472 AssumeClean = 300,
577fd104 473 WriteZeroes,
997aed5d
NB
474 BitmapChunk,
475 WriteBehind,
476 ReAdd,
477 NoDegraded,
478 Sparc22,
1c7a808c 479 BackupFile,
997aed5d 480 HomeHost,
589395d6 481 AutoHomeHost,
1f48664b 482 AutoDetect,
1770662b 483 Waitclean,
4cce4069 484 DetailPlatform,
33414a01 485 KillSubarray,
1c7a808c 486 UpdateSubarray,
edde9560 487 IncrementalPath,
1c7a808c
N
488 NoSharing,
489 HelpOptions,
490 Brief,
d11abe4b 491 NoDevices,
1c7a808c
N
492 ManageOpt,
493 Add,
f33a71f1 494 AddSpare,
01290056 495 AddJournal,
1c7a808c
N
496 Remove,
497 Fail,
70c55e36
N
498 Replace,
499 With,
1c7a808c
N
500 MiscOpt,
501 WaitOpt,
502 ConfigFile,
503 ChunkSize,
504 WriteMostly,
71574efb
N
505 FailFast,
506 NoFailFast,
1c7a808c
N
507 Layout,
508 Auto,
509 Force,
510 SuperMinor,
511 EMail,
512 ProgramOpt,
513 Increment,
514 Fork,
515 Bitmap,
516 RebuildMapOpt,
87f26d14 517 InvalidBackup,
20b60dcd 518 UdevRules,
b76b30e0 519 FreezeReshape,
2dddadb0 520 Continue,
08ca2adf 521 OffRootOpt,
c2ecf5f6 522 Prefer,
f7d3febc 523 KillOpt,
40c9a66a 524 DataOffset,
6d388a88 525 ExamineBB,
74db60b0
N
526 Dump,
527 Restore,
a740cf64 528 Action,
529e2aa5 529 Nodes,
7716570e 530 ClusterName,
4de90913 531 ClusterConfirm,
cc1799c3 532 WriteJournal,
5308f117 533 ConsistencyPolicy,
997aed5d
NB
534};
535
2568ce89
MK
536enum update_opt {
537 UOPT_NAME = 1,
538 UOPT_PPL,
539 UOPT_NO_PPL,
540 UOPT_BITMAP,
541 UOPT_NO_BITMAP,
542 UOPT_SUBARRAY_ONLY,
543 UOPT_SPARC22,
544 UOPT_SUPER_MINOR,
545 UOPT_SUMMARIES,
546 UOPT_RESYNC,
547 UOPT_UUID,
548 UOPT_HOMEHOST,
549 UOPT_HOME_CLUSTER,
550 UOPT_NODES,
551 UOPT_DEVICESIZE,
552 UOPT_BBL,
553 UOPT_NO_BBL,
554 UOPT_FORCE_NO_BBL,
555 UOPT_METADATA,
556 UOPT_REVERT_RESHAPE,
557 UOPT_LAYOUT_ORIGINAL,
558 UOPT_LAYOUT_ALTERNATE,
559 UOPT_LAYOUT_UNSPECIFIED,
560 UOPT_BYTEORDER,
561 UOPT_HELP,
2257de10
MK
562 UOPT_USER_ONLY,
563 /*
564 * Code specific options, cannot be set by the user
565 */
566 UOPT_SPEC_FORCE_ONE,
567 UOPT_SPEC_FORCE_ARRAY,
568 UOPT_SPEC_ASSEMBLE,
569 UOPT_SPEC_LINEAR_GROW_NEW,
570 UOPT_SPEC_LINEAR_GROW_UPDATE,
571 UOPT_SPEC__RESHAPE_PROGRESS,
572 UOPT_SPEC_WRITEMOSTLY,
573 UOPT_SPEC_READWRITE,
574 UOPT_SPEC_FAILFAST,
575 UOPT_SPEC_NOFAILFAST,
576 UOPT_SPEC_REVERT_RESHAPE_NOBACKUP,
2568ce89
MK
577 UOPT_UNDEFINED
578};
579extern void fprint_update_options(FILE *outf, enum update_opt update_mode);
580
f0ec6710
MN
581enum prefix_standard {
582 JEDEC,
583 IEC
584};
585
0aa2f15b
GJ
586enum bitmap_update {
587 NoUpdate,
588 NameUpdate,
7e6e839a 589 NodeNumUpdate,
0aa2f15b
GJ
590};
591
e22fe3ae
N
592enum flag_mode {
593 FlagDefault, FlagSet, FlagClear,
594};
595
64c4757e 596/* structures read from config file */
52826846
NB
597/* List of mddevice names and identifiers
598 * Identifiers can be:
599 * uuid=128-hex-uuid
600 * super-minor=decimal-minor-number-from-superblock
601 * devices=comma,separated,list,of,device,names,with,wildcards
602 *
603 * If multiple fields are present, the intersection of all matching
604 * devices is considered
605 */
98c6faba 606#define UnSet (0xfffe)
fa56eddb 607struct mddev_ident {
dd0781e5 608 char *devname;
aba69144 609
dd0781e5 610 int uuid_set;
3fa06e9d 611 int uuid[4];
67417d92 612 char name[MD_NAME_MAX + 1];
52826846 613
f21e18ca 614 int super_minor;
52826846 615
dd0781e5 616 char *devices; /* comma separated list of device
52826846
NB
617 * names with wild cards
618 */
dd0781e5 619 int level;
f21e18ca
N
620 int raid_disks;
621 int spare_disks;
82d9eba6 622 struct supertype *st;
dd0781e5
NB
623 int autof; /* 1 for normal, 2 for partitioned */
624 char *spare_group;
7ef02d01 625 char *bitmap_file;
c82f047c 626 int bitmap_fd;
dd0781e5 627
1771a6e2
N
628 char *container; /* /dev/whatever name of container, or
629 * uuid of container. You would expect
630 * this to be the 'devname' or UUID
dbb44303
N
631 * of some other entry.
632 */
633 char *member; /* subarray within a container */
634
fa56eddb 635 struct mddev_ident *next;
b179246f
N
636 union {
637 /* fields needed by different users of this structure */
638 int assembled; /* set when assembly succeeds */
639 };
fa56eddb 640};
64c4757e 641
9e33d556
N
642struct context {
643 int readonly;
644 int runstop;
645 int verbose;
9e33d556 646 int brief;
d11abe4b 647 int no_devices;
9e33d556
N
648 int force;
649 char *homehost;
650 int require_homehost;
651 char *prefer;
652 int export;
653 int test;
654 char *subarray;
3a87fa67 655 enum update_opt update;
9e33d556
N
656 int scan;
657 int SparcAdjust;
658 int autof;
659 int delay;
660 int freeze_reshape;
661 char *backup_file;
662 int invalid_backup;
a740cf64 663 char *action;
529e2aa5 664 int nodes;
7716570e 665 char *homecluster;
9e33d556
N
666};
667
e705e81b
N
668struct shape {
669 int raiddisks;
670 int sparedisks;
cc1799c3 671 int journaldisks;
e705e81b
N
672 int level;
673 int layout;
674 char *layout_str;
675 int chunk;
676 int bitmap_chunk;
677 char *bitmap_file;
678 int assume_clean;
577fd104 679 bool write_zeroes;
e705e81b
N
680 int write_behind;
681 unsigned long long size;
ae5dfc56 682 unsigned long long data_offset;
5308f117 683 int consistency_policy;
e705e81b
N
684};
685
64c4757e 686/* List of device names - wildcards expanded */
a655e550 687struct mddev_dev {
64c4757e 688 char *devname;
c8e1a230
N
689 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail,
690 * 'A' for re_add.
cd29a5c8
NB
691 * Not set for names read from .config
692 */
e22fe3ae
N
693 enum flag_mode writemostly;
694 enum flag_mode failfast;
70c55e36 695 int used; /* set when used */
72ca9bcf 696 long long data_offset;
a655e550
N
697 struct mddev_dev *next;
698};
64c4757e 699
682c7051
NB
700typedef struct mapping {
701 char *name;
702 int num;
703} mapping_t;
704
e0d19036 705struct mdstat_ent {
4dd2df09 706 char devnm[32];
e0d19036
NB
707 int active;
708 char *level;
7054da69 709 char *pattern; /* U for up, _ for down */
e0d19036 710 int percent; /* -1 if no resync */
f94c116f 711 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
712 int devcnt;
713 int raid_disks;
549e9569 714 char * metadata_version;
3b57c466
N
715 struct dev_member {
716 char *name;
717 struct dev_member *next;
1011e834 718 } *members;
e0d19036
NB
719 struct mdstat_ent *next;
720};
721
22a88995 722extern struct mdstat_ent *mdstat_read(int hold, int start);
a7a0d8a1 723extern void mdstat_close(void);
e0d19036 724extern void free_mdstat(struct mdstat_ent *ms);
cab9c67d 725extern int mdstat_wait(int seconds);
58a4ba2a 726extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
4dd2df09 727extern int mddev_busy(char *devnm);
3b57c466 728extern struct mdstat_ent *mdstat_by_component(char *name);
4dd2df09 729extern struct mdstat_ent *mdstat_by_subdev(char *subdev, char *container);
8382f19b
NB
730
731struct map_ent {
732 struct map_ent *next;
4dd2df09 733 char devnm[32];
1522c538 734 char metadata[20];
8382f19b 735 int uuid[4];
195254b8 736 int bad;
8382f19b
NB
737 char *path;
738};
4dd2df09 739extern int map_update(struct map_ent **mpp, char *devnm, char *metadata,
8382f19b 740 int uuid[4], char *path);
4dd2df09 741extern void map_remove(struct map_ent **map, char *devnm);
8382f19b 742extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
4dd2df09 743extern struct map_ent *map_by_devnm(struct map_ent **map, char *devnm);
eb2306f8 744extern void map_free(struct map_ent *map);
f2e55ecc 745extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
746extern void map_read(struct map_ent **melp);
747extern int map_write(struct map_ent *mel);
4dd2df09 748extern void map_delete(struct map_ent **mapp, char *devnm);
8382f19b 749extern void map_add(struct map_ent **melp,
4dd2df09 750 char *devnm, char *metadata, int uuid[4], char *path);
ad5bc697
N
751extern int map_lock(struct map_ent **melp);
752extern void map_unlock(struct map_ent **melp);
cc700db3 753extern void map_fork(void);
e0d19036 754
e86c9dd6 755/* various details can be requested */
dab4a513
DW
756enum sysfs_read_flags {
757 GET_LEVEL = (1 << 0),
758 GET_LAYOUT = (1 << 1),
759 GET_COMPONENT = (1 << 2),
760 GET_CHUNK = (1 << 3),
761 GET_CACHE = (1 << 4),
762 GET_MISMATCH = (1 << 5),
763 GET_VERSION = (1 << 6),
764 GET_DISKS = (1 << 7),
dab4a513 765 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
766 GET_BITMAP_LOCATION = (1 << 10),
767
768 GET_DEVS = (1 << 20), /* gets role, major, minor */
769 GET_OFFSET = (1 << 21),
770 GET_SIZE = (1 << 22),
771 GET_STATE = (1 << 23),
772 GET_ERROR = (1 << 24),
5aa644c6 773 GET_ARRAY_STATE = (1 << 25),
5308f117 774 GET_CONSISTENCY_POLICY = (1 << 26),
ae7d61e3 775 GET_DEVS_ALL = (1 << 27),
dab4a513 776};
e86c9dd6 777
90fd7001
MK
778#define SYSFS_MAX_BUF_SIZE 64
779
14a86579
MT
780extern void sysfs_get_container_devnm(struct mdinfo *mdi, char *buf);
781
e86c9dd6 782/* If fd >= 0, get the array it is open on,
4dd2df09 783 * else use devnm.
e86c9dd6 784 */
4dd2df09 785extern int sysfs_open(char *devnm, char *devname, char *attr);
dae13137 786extern int sysfs_init(struct mdinfo *mdi, int fd, char *devnm);
a37563c9 787extern void sysfs_init_dev(struct mdinfo *mdi, dev_t devid);
7e0f6979 788extern void sysfs_free(struct mdinfo *sra);
4dd2df09 789extern struct mdinfo *sysfs_read(int fd, char *devnm, unsigned long options);
1770662b
DW
790extern int sysfs_attr_match(const char *attr, const char *str);
791extern int sysfs_match_word(const char *word, char **list);
7e0f6979 792extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 793 char *name, char *val);
7e0f6979 794extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 795 char *name, unsigned long long val);
012a8641
JS
796extern int sysfs_set_num_signed(struct mdinfo *sra, struct mdinfo *dev,
797 char *name, long long val);
97590376 798extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
799extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
800 char *name);
801extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 802extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 803 char *name, unsigned long long *val);
2eba8496
N
804extern int sysfs_fd_get_two(int fd, unsigned long long *v1, unsigned long long *v2);
805extern int sysfs_get_two(struct mdinfo *sra, struct mdinfo *dev,
806 char *name, unsigned long long *v1, unsigned long long *v2);
7236ee7a 807extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
808extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
809 char *name);
93ecfa01
N
810extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
811 char *name, char *val, int size);
8ed3e5e1 812extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
f35f2525 813extern int sysfs_set_array(struct mdinfo *info, int vers);
2904b26f 814extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 815extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
4dd2df09 816extern int sysfs_unique_holder(char *devnm, long rdev);
bc77ed53 817extern int sysfs_freeze_array(struct mdinfo *sra);
efc67e8e 818extern int sysfs_wait(int fd, int *msec);
193b6c0b 819extern int load_sys(char *path, char *buf, int len);
b2514242 820extern int zero_disk_range(int fd, unsigned long long sector, size_t count);
130994cb
AK
821extern int reshape_prepare_fdlist(char *devname,
822 struct mdinfo *sra,
823 int raid_disks,
824 int nrdisks,
825 unsigned long blocks,
826 char *backup_file,
827 int *fdlist,
828 unsigned long long *offsets);
829extern void reshape_free_fdlist(int *fdlist,
830 unsigned long long *offsets,
831 int size);
e6e9d47b
AK
832extern int reshape_open_backup_file(char *backup,
833 int fd,
834 char *devname,
835 long blocks,
836 int *fdlist,
a93f87ee 837 unsigned long long *offsets,
54ded86f 838 char *sysfs_name,
a93f87ee 839 int restart);
1c009fc2
AK
840extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
841 unsigned int ndata, unsigned int odata);
54ded86f
N
842extern char *locate_backup(char *name);
843extern char *make_backup(char *name);
e86c9dd6
NB
844
845extern int save_stripes(int *source, unsigned long long *offsets,
846 int raid_disks, int chunk_size, int level, int layout,
847 int nwrites, int *dest,
a6288483
N
848 unsigned long long start, unsigned long long length,
849 char *buf);
353632d9
NB
850extern int restore_stripes(int *dest, unsigned long long *offsets,
851 int raid_disks, int chunk_size, int level, int layout,
852 int source, unsigned long long read_offset,
2fcb75ae
AK
853 unsigned long long start, unsigned long long length,
854 char *src_buf);
e86c9dd6 855
52826846
NB
856#ifndef Sendmail
857#define Sendmail "/usr/lib/sendmail -t"
858#endif
859
773135f5 860#define SYSLOG_FACILITY LOG_DAEMON
5f21d674 861extern char *map_num_s(mapping_t *map, int num);
682c7051
NB
862extern char *map_num(mapping_t *map, int num);
863extern int map_name(mapping_t *map, char *name);
329dfc28
N
864extern mapping_t r0layout[], r5layout[], r6layout[],
865 pers[], modes[], faultylayout[];
2568ce89 866extern mapping_t consistency_policies[], sysfs_array_states[], update_options[];
682c7051 867
c2ecf5f6
N
868extern char *map_dev_preferred(int major, int minor, int create,
869 char *prefer);
870static inline char *map_dev(int major, int minor, int create)
871{
872 return map_dev_preferred(major, minor, create, NULL);
873}
64c4757e 874
4389ce73
MT
875/**
876 * is_fd_valid() - check file descriptor.
877 * @fd: file descriptor.
878 *
879 * The function checks if @fd is nonnegative integer and shall be used only
880 * to verify open() result.
881 */
882static inline int is_fd_valid(int fd)
883{
884 return (fd > -1);
885}
886
70f1ff42
LF
887/**
888 * is_level456() - check whether given level is between inclusive 4 and 6.
889 * @level: level to check.
890 *
891 * Return: true if condition is met, false otherwise
892 */
893static inline bool is_level456(int level)
894{
895 return (level >= 4 && level <= 6);
896}
897
4389ce73
MT
898/**
899 * close_fd() - verify, close and unset file descriptor.
900 * @fd: pointer to file descriptor.
901 *
902 * The function closes and invalidates file descriptor if appropriative. It
903 * ignores incorrect file descriptor quitely to simplify error handling.
904 */
905static inline void close_fd(int *fd)
906{
907 if (is_fd_valid(*fd) && close(*fd) == 0)
908 *fd = -1;
909}
910
549e9569 911struct active_array;
2e735d19 912struct metadata_update;
64c4757e 913
80bf9135 914/* 'struct reshape' records the intermediate states of
999b4972
N
915 * a general reshape.
916 * The starting geometry is converted to the 'before' geometry
917 * by at most an atomic level change. They could be the same.
918 * Similarly the 'after' geometry is converted to the final
919 * geometry by at most a level change.
920 * Note that 'before' and 'after' must have the same level.
921 * 'blocks' is the minimum number of sectors for a reshape unit.
922 * This will be a multiple of the stripe size in each of the
923 * 'before' and 'after' geometries.
924 * If 'blocks' is 0, no restriping is necessary.
89ecd3cf
N
925 * 'min_offset_change' is the minimum change to data_offset to
926 * allow the reshape to happen. It is at least the larger of
927 * the old and new chunk sizes, and typically the same as 'blocks'
928 * divided by number of data disks.
999b4972
N
929 */
930struct reshape {
931 int level;
932 int parity; /* number of parity blocks/devices */
933 struct {
934 int layout;
935 int data_disks;
936 } before, after;
937 unsigned long long backup_blocks;
89ecd3cf 938 unsigned long long min_offset_change;
999b4972
N
939 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
940 unsigned long long new_size; /* New size of array in sectors */
941};
942
1fef0c6f
MT
943/**
944 * struct dev_policy - Data structure for policy management.
945 * @next: pointer to next dev_policy.
946 * @name: policy name, category.
947 * @metadata: the metadata type it affects.
948 * @value: value of the policy.
949 *
950 * The functions to manipulate dev_policy lists do not free elements, so they must be statically
951 * allocated. @name and @metadata can be compared by address.
952 */
953typedef struct dev_policy {
954 struct dev_policy *next;
955 char *name;
956 const char *metadata;
957 const char *value;
958} dev_policy_t;
959
7054da69 960/* A superswitch provides entry point to a metadata handler.
6adfd3af 961 *
74db60b0 962 * The superswitch primarily operates on some "metadata" that
6adfd3af
NB
963 * is accessed via the 'supertype'.
964 * This metadata has one of three possible sources.
965 * 1/ It is read from a single device. In this case it may not completely
966 * describe the array or arrays as some information might be on other
967 * devices.
968 * 2/ It is read from all devices in a container. In this case all
969 * information is present.
970 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
971 * be complete once enough ->add_to_super calls have completed.
972 *
973 * When creating an array inside a container, the metadata will be
974 * formed by a combination of 2 and 3. The metadata or the array is read,
975 * then new information is added.
976 *
977 * The metadata must sometimes have a concept of a 'current' array
978 * and a 'current' device.
979 * The 'current' array is set by init_super to be the newly created array,
980 * or is set by super_by_fd when it finds it is looking at an array inside
981 * a container.
982 *
983 * The 'current' device is either the device that the metadata was read from
984 * in case 1, or the last device added by add_to_super in case 3.
985 * Case 2 does not identify a 'current' device.
986 */
f9ce90ba 987extern struct superswitch {
6adfd3af
NB
988
989 /* Used to report details of metadata read from a component
990 * device. ->load_super has been called.
991 */
3da92f27 992 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 993 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 994 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 995 void (*export_examine_super)(struct supertype *st);
6d388a88 996 int (*examine_badblocks)(struct supertype *st, int fd, char *devname);
74db60b0 997 int (*copy_metadata)(struct supertype *st, int from, int to);
6adfd3af
NB
998
999 /* Used to report details of an active array.
1000 * ->load_super was possibly given a 'component' string.
1001 */
b771faef
BK
1002 void (*detail_super)(struct supertype *st, char *homehost,
1003 char *subarray);
1004 void (*brief_detail_super)(struct supertype *st, char *subarray);
0d726f17 1005 void (*export_detail_super)(struct supertype *st);
6adfd3af 1006
4cce4069 1007 /* Optional: platform hardware / firmware details */
9eafa1de
MN
1008 int (*detail_platform)(int verbose, int enumerate_only, char *controller_path);
1009 int (*export_detail_platform)(int verbose, char *controller_path);
4cce4069 1010
6adfd3af
NB
1011 /* Used:
1012 * to get uuid to storing in bitmap metadata
1013 * and 'reshape' backup-data metadata
1014 * To see if a device is being re-added to an array it was part of.
1015 */
3da92f27 1016 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 1017
8592f29d 1018 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
1019 * the container, or about an individual array within the container.
1020 * The determination is made either by:
1021 * load_super being given a 'component' string.
1022 * validate_geometry determining what to create.
d2ca6449
NB
1023 * The info includes both array information and device information.
1024 * The particular device should be:
1025 * The last device added by add_to_super
1026 * The device the metadata was loaded from by load_super
a5d85af7
N
1027 * If 'map' is present, then it is an array raid_disks long
1028 * (raid_disk must already be set and correct) and it is filled
1029 * with 1 for slots that are thought to be active and 0 for slots which
1030 * appear to be failed/missing.
95eeceeb 1031 * *info is zeroed out before data is added.
6adfd3af 1032 */
a5d85af7 1033 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 1034 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 1035 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 1036 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 1037 */
3da92f27 1038 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
1039
1040 /* Make one of several generic modifications to metadata
1041 * prior to assembly (or other times).
1042 * sparc2.2 - first bug in early 0.90 metadata
1043 * super-minor - change name of 0.90 metadata
1044 * summaries - 'correct' any redundant data
1045 * resync - mark array as dirty to trigger a resync.
1046 * uuid - set new uuid - only 0.90 or 1.x
1047 * name - change the name of the array (where supported)
1048 * homehost - change which host this array is tied to.
1049 * devicesize - If metadata is at start of device, change recorded
1050 * device size to match actual device size
1051 * byteorder - swap bytes for 0.90 metadata
1052 *
1053 * force-one - mark that device as uptodate, not old or failed.
1054 * force-array - mark array as clean if it would not otherwise
1055 * assemble
1056 * assemble - not sure how this is different from force-one...
1057 * linear-grow-new - add a new device to a linear array, but don't
1058 * change the size: so superblock still matches
1059 * linear-grow-update - now change the size of the array.
16715c01
DL
1060 * writemostly - set the WriteMostly1 bit in the superblock devflags
1061 * readwrite - clear the WriteMostly1 bit in the superblock devflags
71574efb
N
1062 * failfast - set the FailFast1 bit in the superblock
1063 * nofailfast - clear the FailFast1 bit
199f1a1f
N
1064 * no-bitmap - clear any record that a bitmap is present.
1065 * bbl - add a bad-block-log if possible
6dd16dac
N
1066 * no-bbl - remove any bad-block-log is it is empty.
1067 * force-no-bbl - remove any bad-block-log even if empty.
199f1a1f
N
1068 * revert-reshape - If a reshape is in progress, modify metadata so
1069 * it will resume going in the opposite direction.
6adfd3af 1070 */
68c7d6d7 1071 int (*update_super)(struct supertype *st, struct mdinfo *info,
03312b52 1072 enum update_opt update,
e5eac01f
NB
1073 char *devname, int verbose,
1074 int uuid_set, char *homehost);
6adfd3af
NB
1075
1076 /* Create new metadata for new array as described. This could
1077 * be a new container, or an array in a pre-existing container.
1078 * Also used to zero metadata prior to writing it to invalidate old
1079 * metadata.
1080 */
3da92f27 1081 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
5308f117 1082 struct shape *s, char *name,
83cd1e97
N
1083 char *homehost, int *uuid,
1084 unsigned long long data_offset);
6adfd3af
NB
1085
1086 /* update the metadata to include new device, either at create or
1087 * when hot-adding a spare.
1088 */
f20c3968 1089 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
72ca9bcf
N
1090 int fd, char *devname,
1091 unsigned long long data_offset);
1a64be56
LM
1092 /* update the metadata to delete a device,
1093 * when hot-removing.
1094 */
1095 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
1096
1097 /* Write metadata to one device when fixing problems or adding
1098 * a new device.
1099 */
3da92f27 1100 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
1101
1102 /* Write all metadata for this array.
1103 */
111d01fc 1104 int (*write_init_super)(struct supertype *st);
e2f408a4
N
1105 /* Check if metadata read from one device is compatible with an array,
1106 * used when assembling an array, or pseudo-assembling was with
1107 * "--examine --brief"
1108 * If "st" has not yet been loaded the superblock from, "tst" is
1109 * moved in, otherwise the superblock in 'st' is compared with
1110 * 'tst'.
1111 */
c7b8547c
MT
1112 int (*compare_super)(struct supertype *st, struct supertype *tst,
1113 int verbose);
e2f408a4
N
1114 /* Load metadata from a single device. If 'devname' is not NULL
1115 * print error messages as appropriate */
3da92f27 1116 int (*load_super)(struct supertype *st, int fd, char *devname);
e2f408a4
N
1117 /* 'fd' is a 'container' md array - load array metadata from the
1118 * whole container.
1119 */
2b959fbf 1120 int (*load_container)(struct supertype *st, int fd, char *devname);
e2f408a4
N
1121 /* If 'arg' is a valid name of this metadata type, allocate and
1122 * return a 'supertype' for the particular minor version */
82d9eba6 1123 struct supertype * (*match_metadata_desc)(char *arg);
e2f408a4
N
1124 /* If a device has the given size, and the data_offset has been
1125 * requested - work out how much space is available for data.
1126 * This involves adjusting for reserved space (e.g. bitmaps)
1127 * and for any rounding.
1128 * 'mdadm' only calls this for existing arrays where a possible
1129 * spare is being added. However some super-handlers call it
1130 * internally from validate_geometry when creating an array.
1131 */
387fcd59
N
1132 __u64 (*avail_size)(struct supertype *st, __u64 size,
1133 unsigned long long data_offset);
fbfdcb06
AO
1134 /*
1135 * Return spare criteria for array:
1136 * - minimum disk size can be used in array;
4b57ecf6 1137 * - sector size can be used in array.
e2f408a4 1138 */
f6562011
MT
1139 mdadm_status_t (*get_spare_criteria)(struct supertype *st, char *mddev_path,
1140 struct spare_criteria *sc);
e2f408a4
N
1141 /* Find somewhere to put a bitmap - possibly auto-size it - and
1142 * update the metadata to record this. The array may be newly
1143 * created, in which case data_size may be updated, or it might
1144 * already exist. Metadata handler can know if init_super
1145 * has been called, but not write_init_super.
2ec2b7e9
JS
1146 * 0: Success
1147 * -Exxxx: On error
e2f408a4 1148 */
3da92f27 1149 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 1150 int delay, int write_behind,
f9c25f1d 1151 unsigned long long size, int may_change, int major);
b554ab5c
JR
1152 /* Perform additional setup required to activate a bitmap.
1153 */
1154 int (*set_bitmap)(struct supertype *st, struct mdinfo *info);
e2f408a4
N
1155 /* Seek 'fd' to start of write-intent-bitmap. Must be an
1156 * md-native format bitmap
1157 */
b138214f 1158 int (*locate_bitmap)(struct supertype *st, int fd, int node_num);
e2f408a4
N
1159 /* if add_internal_bitmap succeeded for existing array, this
1160 * writes it out.
1161 */
0aa2f15b 1162 int (*write_bitmap)(struct supertype *st, int fd, enum bitmap_update update);
e2f408a4 1163 /* Free the superblock and any other allocated data */
3da92f27 1164 void (*free_super)(struct supertype *st);
78e44928
NB
1165
1166 /* validate_geometry is called with an st returned by
1167 * match_metadata_desc.
e2f408a4 1168 * It should check that the geometry described is compatible with
78e44928
NB
1169 * the metadata type. It will be called repeatedly as devices
1170 * added to validate changing size and new devices. If there are
1171 * inter-device dependencies, it should record sufficient details
1172 * so these can be validated.
30f58b22 1173 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
1174 * Return value is:
1175 * 1: everything is OK
1176 * 0: not OK for some reason - if 'verbose', then error was reported.
1177 * -1: st->sb was NULL, 'subdev' is a member of a container of this
e2f408a4 1178 * type, but array is not acceptable for some reason
ecbd9e81 1179 * message was reported even if verbose is 0.
78e44928 1180 */
17f25ca6
NB
1181 int (*validate_geometry)(struct supertype *st, int level, int layout,
1182 int raiddisks,
c21e737b 1183 int *chunk, unsigned long long size,
af4348dd 1184 unsigned long long data_offset,
2c514b71 1185 char *subdev, unsigned long long *freesize,
5308f117 1186 int consistency_policy, int verbose);
598f0d58 1187
1fef0c6f
MT
1188 /**
1189 * test_and_add_drive_policies() - test new and add custom policies from metadata handler.
1190 * @pols: list of currently recorded policies.
1191 * @disk_fd: file descriptor of the device to check.
1192 * @verbose: verbose flag.
1193 *
1194 * Used by IMSM to verify all drives in container/array, against requirements not recored
1195 * in superblock, like controller type for IMSM. It should check all drives even if
1196 * they are not actually used, because mdmon or kernel are free to use any drive assigned to
1197 * container automatically.
1198 *
1199 * Generating and comparison methods belong to metadata handler. It is not mandatory to be
1200 * implemented.
1201 *
1202 * Return: MDADM_STATUS_SUCCESS is expected on success.
1203 */
1204 mdadm_status_t (*test_and_add_drive_policies)(dev_policy_t **pols, int disk_fd,
1205 const int verbose);
1206
e2f408a4
N
1207 /* Return a linked list of 'mdinfo' structures for all arrays
1208 * in the container. For non-containers, it is like
1209 * getinfo_super with an allocated mdinfo.*/
00bbdbda 1210 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
1211 /* query the supertype for default geometry */
1212 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01 1213 /* Permit subarray's to be deleted from inactive containers */
3364781b
BK
1214 int (*kill_subarray)(struct supertype *st,
1215 char *subarray_id); /* optional */
03312b52
MK
1216 /**
1217 * update_subarray() - Permit subarray to be modified.
1218 * @st: Supertype.
1219 * @subarray: Subarray name.
1220 * @update: Update option.
1221 * @ident: Optional identifiers.
1222 */
a951a4f7 1223 int (*update_subarray)(struct supertype *st, char *subarray,
03312b52 1224 enum update_opt update, struct mddev_ident *ident);
7bc71196
DW
1225 /* Check if reshape is supported for this external format.
1226 * st is obtained from super_by_fd() where st->subarray[0] is
1227 * initialized to indicate if reshape is being performed at the
1228 * container or subarray level
1229 */
016e00f5
AK
1230#define APPLY_METADATA_CHANGES 1
1231#define ROLLBACK_METADATA_CHANGES 0
1232
d04f65f4
N
1233 int (*reshape_super)(struct supertype *st,
1234 unsigned long long size, int level,
7bc71196 1235 int layout, int chunksize, int raid_disks,
41784c88 1236 int delta_disks, char *backup, char *dev,
016e00f5 1237 int direction,
41784c88 1238 int verbose); /* optional */
999b4972
N
1239 int (*manage_reshape)( /* optional */
1240 int afd, struct mdinfo *sra, struct reshape *reshape,
1241 struct supertype *st, unsigned long blocks,
1242 int *fds, unsigned long long *offsets,
1243 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 1244
549e9569 1245/* for mdmon */
cba0191b 1246 int (*open_new)(struct supertype *c, struct active_array *a,
60815698 1247 int inst);
ed9d66aa
NB
1248
1249 /* Tell the metadata handler the current state of the array.
1250 * This covers whether it is known to be consistent (no pending writes)
01f157d7 1251 * and how far along a resync is known to have progressed
ed9d66aa
NB
1252 * (in a->resync_start).
1253 * resync status is really irrelevant if the array is not consistent,
1254 * but some metadata (DDF!) have a place to record the distinction.
1011e834 1255 * If 'consistent' is '2', then the array can mark it dirty if a
01f157d7
N
1256 * resync/recovery/whatever is required, or leave it clean if not.
1257 * Return value is 0 dirty (not consistent) and 1 if clean.
1258 * it is only really important if consistent is passed in as '2'.
4e5528c6 1259 */
01f157d7 1260 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
1261
1262 /* When the state of a device might have changed, we call set_disk to
1263 * tell the metadata what the current state is.
1264 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
1265 * transitions.
1266 * set_disk might be called when the state of the particular disk has
1267 * not in fact changed.
1268 */
8d45d196 1269 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
1270 void (*sync_metadata)(struct supertype *st);
1271 void (*process_update)(struct supertype *st,
1272 struct metadata_update *update);
5fe6f031
N
1273 /* Prepare updates allocates extra memory that might be
1274 * needed. If the update cannot be understood, return 0.
1275 */
1276 int (*prepare_update)(struct supertype *st,
edd8d13c 1277 struct metadata_update *update);
549e9569 1278
6c3fb95c
NB
1279 /* activate_spare will check if the array is degraded and, if it
1280 * is, try to find some spare space in the container.
1281 * On success, it add appropriate updates (For process_update) to
1282 * to the 'updates' list and returns a list of 'mdinfo' identifying
1283 * the device, or devices as there might be multiple missing
1284 * devices and multiple spares available.
1285 */
1286 struct mdinfo *(*activate_spare)(struct active_array *a,
1287 struct metadata_update **updates);
2cda7640
ML
1288 /*
1289 * Return statically allocated string that represents metadata specific
1290 * controller domain of the disk. The domain is used in disk domain
1291 * matching functions. Disks belong to the same domain if the they have
1292 * the same domain from mdadm.conf and belong the same metadata domain.
1293 * Returning NULL or not providing this handler means that metadata
1294 * does not distinguish the differences between disks that belong to
1295 * different controllers. They are in the domain specified by
1296 * configuration file (mdadm.conf).
1297 * In case when the metadata has the notion of domains based on disk
1298 * it shall return NULL for disks that do not belong to the controller
1299 * the supported domains. Such disks will form another domain and won't
1300 * be mixed with supported ones.
1301 */
1302 const char *(*get_disk_controller_domain)(const char *path);
549e9569 1303
ba53ea59
AK
1304 /* for external backup area */
1305 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
1306
0c21b485
PB
1307 /* validate container after assemble */
1308 int (*validate_container)(struct mdinfo *info);
1309
5308f117
AP
1310 /* write initial empty PPL on device */
1311 int (*write_init_ppl)(struct supertype *st, struct mdinfo *info, int fd);
1312
2432ce9b
AP
1313 /* validate ppl before assemble */
1314 int (*validate_ppl)(struct supertype *st, struct mdinfo *info,
1315 struct mdinfo *disk);
1316
1ab97c97
TM
1317 /* records new bad block in metadata */
1318 int (*record_bad_block)(struct active_array *a, int n,
1319 unsigned long long sector, int length);
1320
42d902d9
TM
1321 /* clears bad block from metadata */
1322 int (*clear_bad_block)(struct active_array *a, int n,
1323 unsigned long long sector, int length);
1324
1325 /* get list of bad blocks from metadata */
1326 struct md_bb *(*get_bad_blocks)(struct active_array *a, int n);
1327
f277ce36 1328 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 1329 int external;
4cce4069 1330 const char *name; /* canonical metadata name */
0f22b998 1331} *superlist[];
82d9eba6 1332
0f22b998
N
1333extern struct superswitch super0, super1;
1334extern struct superswitch super_imsm, super_ddf;
0592faeb 1335extern struct superswitch mbr, gpt;
cdddbdbc 1336
420dafcd
N
1337void imsm_set_no_platform(int v);
1338
edd8d13c
NB
1339struct metadata_update {
1340 int len;
1341 char *buf;
1342 void *space; /* allocated space that monitor will use */
cb23f1f4
N
1343 void **space_list; /* list of allocated spaces that monitor can
1344 * use or that it returned.
1345 */
edd8d13c
NB
1346 struct metadata_update *next;
1347};
1348
6adfd3af
NB
1349/* A supertype holds a particular collection of metadata.
1350 * It identifies the metadata type by the superswitch, and the particular
1351 * sub-version of that metadata type.
1352 * metadata read in or created is stored in 'sb' and 'info'.
1353 * There are also fields used by mdmon to track containers.
1354 *
352452c3
N
1355 * A supertype may refer to:
1356 * Just an array, possibly in a container
1357 * A container, not identifying any particular array
1358 * Info read from just one device, not yet fully describing the array/container.
1359 *
1360 *
6adfd3af
NB
1361 * A supertype is created by:
1362 * super_by_fd
1363 * guess_super
1364 * dup_super
1365 */
82d9eba6
NB
1366struct supertype {
1367 struct superswitch *ss;
1368 int minor_version;
ea329559 1369 int max_devs;
4dd2df09 1370 char container_devnm[32]; /* devnm of container */
64557c33 1371 void *sb;
111d01fc 1372 void *info;
afa368f4
N
1373 void *other; /* Hack used to convert v0.90 to v1.0 */
1374 unsigned long long devsize;
23bf42cc 1375 unsigned long long data_offset; /* used by v1.x only */
df3346e6
LM
1376 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
1377 HW/firmware related incompatability to load metadata.
1378 Used when examining metadata to display content of disk
1379 when user has no hw/firmare compatible system.
1380 */
edd8d13c
NB
1381 struct metadata_update *updates;
1382 struct metadata_update **update_tail;
1383
549e9569
NB
1384 /* extra stuff used by mdmon */
1385 struct active_array *arrays;
549e9569 1386 int sock; /* listen to external programs */
4dd2df09
N
1387 char devnm[32]; /* e.g. md0. This appears in metadata_version:
1388 * external:/md0/12
1389 */
549e9569 1390 int devcnt;
68226a80 1391 int retry_soon;
529e2aa5 1392 int nodes;
7716570e 1393 char *cluster_name;
549e9569
NB
1394
1395 struct mdinfo *devs;
1396
82d9eba6 1397};
f9ce90ba 1398
4725bc31 1399extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
1400enum guess_types { guess_any, guess_array, guess_partitions };
1401extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
1402static inline struct supertype *guess_super(int fd) {
1403 return guess_super_type(fd, guess_any);
1404}
3da92f27 1405extern struct supertype *dup_super(struct supertype *st);
beae1dfe 1406extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
32971509 1407extern int get_dev_sector_size(int fd, char *dname, unsigned int *sectsizep);
3a371610 1408extern int must_be_container(int fd);
a7c6e3fb 1409void wait_for(char *dev, int fd);
f9ce90ba 1410
4e8d9f0a 1411extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
1412
1413/* iterate over the sublist starting at list, having the same
1414 * 'name' as 'list', and matching the given metadata (Where
1415 * NULL matches anything
1416 */
1417#define pol_for_each(item, list, _metadata) \
1418 for (item = list; \
1419 item && item->name == list->name; \
1420 item = item->next) \
1421 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
1422 ; else
1423
1424/*
1425 * policy records read from mdadm are largely just name-value pairs.
1426 * The names are constants, not strdupped
1427 */
1428struct pol_rule {
1429 struct pol_rule *next;
1430 char *type; /* rule_policy or rule_part */
1431 struct rule {
1432 struct rule *next;
1433 char *name;
1434 char *value;
1435 char *dups; /* duplicates of 'value' with a partNN appended */
1436 } *rule;
1437};
1438
1439extern char rule_policy[], rule_part[];
1440extern char rule_path[], rule_type[];
1441extern char type_part[], type_disk[];
1442
1443extern void policyline(char *line, char *type);
4e8d9f0a 1444extern void policy_add(char *type, ...);
5527fc74
N
1445extern void policy_free(void);
1446
cd72f9d1 1447extern struct dev_policy *path_policy(char **paths, char *type);
5527fc74 1448extern struct dev_policy *disk_policy(struct mdinfo *disk);
4dd2df09 1449extern struct dev_policy *devid_policy(int devid);
5527fc74
N
1450extern void dev_policy_free(struct dev_policy *p);
1451
e78dda3b 1452extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
1453extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
1454
5a2e194c
MT
1455extern mdadm_status_t drive_test_and_add_policies(struct supertype *st, dev_policy_t **pols,
1456 int fd, const int verbose);
1457extern mdadm_status_t sysfs_test_and_add_drive_policies(struct supertype *st, dev_policy_t **pols,
1458 struct mdinfo *mdi, const int verbose);
1459extern mdadm_status_t mddev_test_and_add_drive_policies(struct supertype *st, dev_policy_t **pols,
1460 int array_fd, const int verbose);
1461
e3bb5f14
N
1462enum policy_action {
1463 act_default,
1464 act_include,
1465 act_re_add,
d2db3045
N
1466 act_spare, /* This only applies to bare devices */
1467 act_spare_same_slot, /* this allows non-bare devices,
1468 * but only if recent removal */
1469 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
1470 act_err
1471};
1472
1473extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
1474 enum policy_action want);
1475extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
1476 enum policy_action want);
f5f12c84
N
1477
1478struct domainlist {
1479 struct domainlist *next;
2cda7640 1480 const char *dom;
f5f12c84
N
1481};
1482
1483extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
1484 const char *metadata);
1485extern struct domainlist *domain_from_array(struct mdinfo *mdi,
1486 const char *metadata);
4dd2df09 1487extern void domainlist_add_dev(struct domainlist **dom, int devid,
e78dda3b 1488 const char *metadata);
f5f12c84
N
1489extern void domain_free(struct domainlist *dl);
1490extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
1491 const char *metadata);
e78dda3b 1492void domain_add(struct domainlist **domp, char *domain);
f5f12c84 1493
403410eb
PC
1494extern void policy_save_path(char *id_path, struct map_ent *array);
1495extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
1496
b0681598
MD
1497extern void sysfs_rules_apply(char *devnm, struct mdinfo *dev);
1498extern void sysfsline(char *line);
1499
838acbc2
NB
1500#if __GNUC__ < 3
1501struct stat64;
1502#endif
1503
45e878bb
NB
1504#define HAVE_NFTW we assume
1505#define HAVE_FTW
1506
2df1f269 1507#ifdef __UCLIBC__
45e878bb 1508# include <features.h>
2df1f269
BRF
1509# ifndef __UCLIBC_HAS_LFS__
1510# define lseek64 lseek
1511# endif
45e878bb
NB
1512# ifndef __UCLIBC_HAS_FTW__
1513# undef HAVE_FTW
1514# undef HAVE_NFTW
1515# endif
1516#endif
1517
1518#ifdef __dietlibc__
1519# undef HAVE_NFTW
1520#endif
1521
e39b673e 1522#if defined(__KLIBC__)
1523# undef HAVE_NFTW
1524# undef HAVE_FTW
1525#endif
1526
45e878bb 1527#ifndef HAVE_NFTW
173fc515 1528# define FTW_PHYS 1
45e878bb
NB
1529# ifndef HAVE_FTW
1530 struct FTW {};
173fc515
NB
1531# endif
1532#endif
1533
45e878bb
NB
1534#ifdef HAVE_FTW
1535# include <ftw.h>
1536#endif
1537
173fc515
NB
1538extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1539
64c4757e 1540extern int Manage_ro(char *devname, int fd, int readonly);
d5a40416 1541extern int Manage_run(char *devname, int fd, struct context *c);
fe7e0e64
N
1542extern int Manage_stop(char *devname, int fd, int quiet,
1543 int will_retry);
64c4757e 1544extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1545 struct mddev_dev *devlist, int verbose, int test,
f2e8393b 1546 enum update_opt update, int force);
1f48664b 1547extern int autodetect(void);
e5329c37 1548extern int Grow_Add_device(char *devname, int fd, char *newdev);
50f01ba5
N
1549extern int Grow_addbitmap(char *devname, int fd,
1550 struct context *c, struct shape *s);
32754b7d 1551extern int Grow_reshape(char *devname, int fd,
e2e53a2d 1552 struct mddev_dev *devlist,
32754b7d 1553 struct context *c, struct shape *s);
06b0d786 1554extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1555 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1556extern int Grow_continue(int mdfd, struct supertype *st,
b76b30e0 1557 struct mdinfo *info, char *backup_file,
06e293d0 1558 int forked, int freeze_reshape);
860f11ed
AP
1559extern int Grow_consistency_policy(char *devname, int fd,
1560 struct context *c, struct shape *s);
b76b30e0 1561
3f54bd62
AK
1562extern int restore_backup(struct supertype *st,
1563 struct mdinfo *content,
1564 int working_disks,
1565 int spares,
54ded86f 1566 char **backup_filep,
3f54bd62 1567 int verbose);
2dddadb0
AK
1568extern int Grow_continue_command(char *devname, int fd,
1569 char *backup_file, int verbose);
64c4757e 1570
7f91af49 1571extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1572 struct mddev_ident *ident,
87f26d14 1573 struct mddev_dev *devlist,
4977146a 1574 struct context *c);
64c4757e 1575
330c07f8
MT
1576extern int Build(struct mddev_ident *ident, struct mddev_dev *devlist, struct shape *s,
1577 struct context *c);
1578
1579extern int Create(struct supertype *st, struct mddev_ident *ident, int subdevs,
1580 struct mddev_dev *devlist, struct shape *s, struct context *c);
64c4757e 1581
ef898ce6 1582extern int Detail(char *dev, struct context *c);
9eafa1de 1583extern int Detail_Platform(struct superswitch *ss, int scan, int verbose, int export, char *controller_path);
e0d19036 1584extern int Query(char *dev);
6d388a88 1585extern int ExamineBadblocks(char *devname, int brief, struct supertype *forcest);
eec3f887
N
1586extern int Examine(struct mddev_dev *devlist, struct context *c,
1587 struct supertype *forcest);
a655e550 1588extern int Monitor(struct mddev_dev *devlist,
e0d19036 1589 char *mailaddr, char *alert_cmd,
95c50205
N
1590 struct context *c,
1591 int daemonise, int oneshot,
1592 int dosyslog, char *pidfile, int increments,
1593 int share);
64c4757e 1594
ba728be7
N
1595extern int Kill(char *dev, struct supertype *st, int force, int verbose, int noexcl);
1596extern int Kill_subarray(char *dev, char *subarray, int verbose);
f2e8393b 1597extern int Update_subarray(char *dev, char *subarray, enum update_opt update, struct mddev_ident *ident, int quiet);
b90c0e9a 1598extern int Wait(char *dev);
1ea04629 1599extern int WaitClean(char *dev, int verbose);
a740cf64 1600extern int SetAction(char *dev, char *action);
9a9dab36 1601
b11fe74d 1602extern int Incremental(struct mddev_dev *devlist, struct context *c,
11b6d91d 1603 struct supertype *st);
8382f19b 1604extern void RebuildMap(void);
d5a40416 1605extern int IncrementalScan(struct context *c, char *devnm);
950bc344 1606extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1607extern int CreateBitmap(char *filename, int force, char uuid[16],
1608 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1609 unsigned long write_behind,
dcec9ee5
NB
1610 unsigned long long array_size,
1611 int major);
55935d51 1612extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
9c030dad 1613extern int IsBitmapDirty(char *filename);
20b60dcd 1614extern int Write_rules(char *rule_name);
bf4fb153 1615extern int bitmap_update_uuid(int fd, int *uuid, int swap);
561ad559
AP
1616
1617/* calculate the size of the bitmap given the array size and bitmap chunksize */
1618static inline unsigned long long
1619bitmap_bits(unsigned long long array_size, unsigned long chunksize)
1620{
1621 return (array_size * 512 + chunksize - 1) / chunksize;
1622}
1623
74db60b0
N
1624extern int Dump_metadata(char *dev, char *dir, struct context *c,
1625 struct supertype *st);
1626extern int Restore_metadata(char *dev, char *dir, struct context *c,
1627 struct supertype *st, int only);
c82f047c 1628
9db2ab4e 1629int md_array_valid(int fd);
3ab8f4bf 1630int md_array_active(int fd);
a822017f 1631int md_array_is_active(struct mdinfo *info);
9cd39f01 1632int md_get_array_info(int fd, struct mdu_array_info_s *array);
018a4882 1633int md_set_array_info(int fd, struct mdu_array_info_s *array);
d97572f5 1634int md_get_disk_info(int fd, struct mdu_disk_info_s *disk);
e0d19036 1635extern int get_linux_version(void);
bc77ed53 1636extern int mdadm_version(char *version);
51d4261c 1637extern unsigned long long parse_size(char *size);
64c4757e 1638extern int parse_uuid(char *str, int uuid[4]);
913f07d1 1639int default_layout(struct supertype *st, int level, int verbose);
5339f996 1640extern int is_near_layout_10(int layout);
4a06e2c2
N
1641extern int parse_layout_10(char *layout);
1642extern int parse_layout_faulty(char *layout);
25aa7329 1643extern int parse_num(int *dest, const char *num);
4de90913 1644extern int parse_cluster_confirm_arg(char *inp, char **devname, int *slot);
682c7051
NB
1645extern int check_ext2(int fd, char *name);
1646extern int check_reiser(int fd, char *name);
1647extern int check_raid(int fd, char *name);
53ed6ac3
KW
1648extern int check_partitions(int fd, char *dname,
1649 unsigned long long freesize,
1650 unsigned long long size);
e2eb503b 1651extern bool is_name_posix_compatible(const char *path);
0a6bff09 1652extern int fstat_is_blkdev(int fd, char *devname, dev_t *rdev);
9e04ac1c 1653extern int stat_is_blkdev(char *devname, dev_t *rdev);
64c4757e 1654
67417d92 1655extern bool is_string_lq(const char * const str, size_t max_len);
a35aa68f 1656extern bool is_dev_alive(char *path);
dd0781e5 1657extern int get_mdp_major(void);
cc742d38 1658extern int get_maj_min(char *dev, int *major, int *minor);
7211116c 1659extern bool is_bit_set(int *val, unsigned char index);
8b0dabea 1660extern int dev_open(char *dev, int flags);
4dd2df09 1661extern int open_dev(char *devnm);
8832342d 1662extern void reopen_mddev(int mdfd);
4dd2df09
N
1663extern int open_dev_flags(char *devnm, int flags);
1664extern int open_dev_excl(char *devnm);
8d80900b 1665extern int is_standard(char *dev, int *nump);
9008ed1c 1666extern int same_dev(char *one, char *two);
9eafa1de 1667extern int compare_paths (char* path1,char* path2);
a7dec3fd 1668extern void enable_fds(int devices);
ff6bb131 1669extern void manage_fork_fds(int close_all);
723d1df4 1670extern int continue_via_systemd(char *devnm, char *service_name, char *prefix);
dd0781e5 1671
7fcbfd7c 1672extern void ident_init(struct mddev_ident *ident);
ae5f13a9 1673extern mdadm_status_t ident_set_devname(struct mddev_ident *ident, const char *devname);
67417d92 1674extern mdadm_status_t ident_set_name(struct mddev_ident *ident, const char *name);
7fcbfd7c 1675
f1ae21c4 1676extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1677extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1678extern struct mddev_dev *conf_get_devs(void);
8382f19b 1679extern int conf_test_dev(char *devname);
4e8d9f0a 1680extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1681extern struct createinfo *conf_get_create_info(void);
1682extern void set_conffile(char *file);
1683extern char *conf_get_mailaddr(void);
1684extern char *conf_get_mailfrom(void);
1685extern char *conf_get_program(void);
0ac91628 1686extern char *conf_get_homehost(int *require_homehostp);
7716570e 1687extern char *conf_get_homecluster(void);
0663137c 1688extern int conf_get_monitor_delay(void);
e0d19036 1689extern char *conf_line(FILE *file);
98c6faba 1690extern char *conf_word(FILE *file, int allow_key);
7103b9b8 1691extern void print_quoted(char *str);
06d2ffc3 1692extern int use_udev(void);
9935cf0f 1693extern void print_escape(char *str);
2eba8496 1694extern unsigned long GCD(unsigned long a, unsigned long b);
0ac91628 1695extern int conf_name_is_free(char *name);
ae5f13a9 1696extern bool is_devname_ignore(const char *devname);
25aa7329
MT
1697extern bool is_devname_md_numbered(const char *devname);
1698extern bool is_devname_md_d_numbered(const char *devname);
7c336758 1699extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1700extern int devname_matches(char *name, char *match);
2244d1a9
N
1701extern struct mddev_ident *conf_match(struct supertype *st,
1702 struct mdinfo *info,
1703 char *devname,
1704 int verbose, int *rvp);
0ac91628 1705
e0d19036
NB
1706extern void free_line(char *line);
1707extern int match_oneof(char *devices, char *devname);
e0d19036 1708extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1709extern const int uuid_zero[4];
f277ce36 1710extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1711extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1712extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
d7288ddc 1713extern char *fname_from_uuid(struct supertype *st,
ff54de6e 1714 struct mdinfo *info, char *buf, char sep);
4b1ac34b 1715extern unsigned long calc_csum(void *super, int bytes);
583315d9 1716extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1717 char *avail);
e0d19036 1718extern int ask(char *mesg);
8fac0577 1719extern unsigned long long get_component_size(int fd);
0430ed48 1720extern void remove_partitions(int fd);
691c6ee1 1721extern int test_partition(int fd);
bfd76b93 1722extern int test_partition_from_id(dev_t id);
577e8448 1723extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1724extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1725 int chunksize, unsigned long long devsize);
edd8d13c
NB
1726extern int flush_metadata_updates(struct supertype *st);
1727extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c 1728extern int assemble_container_content(struct supertype *st, int mdfd,
11b6d91d
N
1729 struct mdinfo *content,
1730 struct context *c,
9ca39acb
N
1731 char *chosen_name, int *result);
1732#define INCR_NO 1
1733#define INCR_UNSAFE 2
1734#define INCR_ALREADY 4
1735#define INCR_YES 8
0c0f09cb
MT
1736
1737extern bool devid_matches_criteria(dev_t devid, struct spare_criteria *sc);
1738extern bool disk_fd_matches_criteria(int disk_fd, struct spare_criteria *sc);
326727d9 1739extern struct mdinfo *container_choose_spares(struct supertype *st,
fbfdcb06 1740 struct spare_criteria *criteria,
326727d9
AC
1741 struct domainlist *domlist,
1742 char *spare_group,
1743 const char *metadata, int get_one);
d52bb542 1744extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1745extern int add_disk(int mdfd, struct supertype *st,
1746 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1747extern int remove_disk(int mdfd, struct supertype *st,
1748 struct mdinfo *sra, struct mdinfo *info);
1ab9ed2a
N
1749extern int hot_remove_disk(int mdfd, unsigned long dev, int force);
1750extern int sys_hot_remove_disk(int statefd, int force);
f35f2525 1751extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1752unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1753
56eb10c0 1754extern char *human_size(long long bytes);
f0ec6710 1755extern char *human_size_brief(long long bytes, int prefix);
e4965ef8 1756extern void print_r10_layout(int layout);
e0d19036 1757
4dd2df09 1758extern char *find_free_devnm(int use_partitions);
63152c1b 1759
e0d19036 1760extern void put_md_name(char *name);
d3c40fab 1761extern char *devid2kname(dev_t devid);
13db17bd
ML
1762extern char *devid2devnm(dev_t devid);
1763extern dev_t devnm2devid(char *devnm);
4dd2df09 1764extern char *get_md_name(char *devnm);
e0d19036 1765
c913b90e 1766extern char DefaultConfFile[];
b5e64645 1767
69207ff6 1768extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
cd6cbb08 1769 char *chosen, int block_udev);
69207ff6
N
1770/* values for 'trustworthy' */
1771#define LOCAL 1
d1d3482b 1772#define LOCAL_ANY 10
69207ff6
N
1773#define FOREIGN 2
1774#define METADATA 3
6be1d39d 1775extern int open_mddev(char *dev, int report_errors);
e702f392 1776extern int is_mddev(char *dev);
a322f70c 1777extern int open_container(int fd);
4dd2df09
N
1778extern int metadata_container_matches(char *metadata, char *devnm);
1779extern int metadata_subdev_matches(char *metadata, char *devnm);
33414a01
DW
1780extern int is_container_member(struct mdstat_ent *ent, char *devname);
1781extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1782extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1783extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1784
4dd2df09
N
1785extern int mdmon_running(char *devnm);
1786extern int mdmon_pid(char *devnm);
40ebbb9c 1787extern int check_env(char *name);
148acb7b 1788extern __u32 random32(void);
c5f71c24 1789extern void random_uuid(__u8 *buf);
4dd2df09 1790extern int start_mdmon(char *devnm);
a931db9e 1791
999b4972
N
1792extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1793 struct supertype *st, unsigned long stripes,
1794 int *fds, unsigned long long *offsets,
1795 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1796void abort_reshape(struct mdinfo *sra);
999b4972 1797
afa368f4
N
1798void *super1_make_v0(struct supertype *st, struct mdinfo *info, mdp_super_t *sb0);
1799
f96b1302
AP
1800extern char *stat2kname(struct stat *st);
1801extern char *fd2kname(int fd);
4dd2df09 1802extern char *stat2devnm(struct stat *st);
27ad4900 1803bool stat_is_md_dev(struct stat *st);
4dd2df09 1804extern char *fd2devnm(int fd);
a931db9e 1805
a9c15847 1806extern int in_initrd(void);
d15a1f72 1807
e80357f8
GJ
1808struct cmap_hooks {
1809 void *cmap_handle; /* corosync lib related */
1810
1811 int (*initialize)(cmap_handle_t *handle);
1812 int (*get_string)(cmap_handle_t handle,
1813 const char *string,
1814 char **name);
1815 int (*finalize)(cmap_handle_t handle);
1816};
1817
1818extern void set_cmap_hooks(void);
1819extern void set_hooks(void);
1820
d15a1f72
GJ
1821struct dlm_hooks {
1822 void *dlm_handle; /* dlm lib related */
1823
1824 dlm_lshandle_t (*create_lockspace)(const char *name,
1825 unsigned int mode);
1b7eb962 1826 dlm_lshandle_t (*open_lockspace)(const char *name);
d15a1f72
GJ
1827 int (*release_lockspace)(const char *name, dlm_lshandle_t ls,
1828 int force);
1829 int (*ls_lock)(dlm_lshandle_t lockspace, uint32_t mode,
1830 struct dlm_lksb *lksb, uint32_t flags,
1831 const void *name, unsigned int namelen,
1832 uint32_t parent, void (*astaddr) (void *astarg),
1833 void *astarg, void (*bastaddr) (void *astarg),
1834 void *range);
1b7eb962
GJ
1835 int (*ls_unlock_wait)(dlm_lshandle_t lockspace, uint32_t lkid,
1836 uint32_t flags, struct dlm_lksb *lksb);
d15a1f72
GJ
1837 int (*ls_get_fd)(dlm_lshandle_t ls);
1838 int (*dispatch)(int fd);
1839};
1840
7716570e 1841extern int get_cluster_name(char **name);
81a8a694 1842extern int dlm_funs_ready(void);
1b7eb962
GJ
1843extern int cluster_get_dlmlock(void);
1844extern int cluster_release_dlmlock(void);
d15a1f72 1845extern void set_dlm_hooks(void);
a9c15847 1846
239b3cc0
MG
1847#define MSEC_TO_NSEC(msec) ((msec) * 1000000)
1848#define USEC_TO_NSEC(usec) ((usec) * 1000)
1849extern void sleep_for(unsigned int sec, long nsec, bool wake_after_interrupt);
ee9dcf95
MG
1850extern bool is_directory(const char *path);
1851extern bool is_file(const char *path);
21e622f2 1852extern int s_gethostname(char *buf, int buf_len);
239b3cc0 1853
2d762ade 1854#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1855#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1856#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1857 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1858
3c558363
N
1859static inline int is_subarray(char *vers)
1860{
e9dd1598 1861 /* The version string for a 'subarray' (an array in a container)
1011e834 1862 * is
e9dd1598 1863 * /containername/componentname for normal read-write arrays
746a6567
N
1864 * -containername/componentname for arrays which mdmon must not
1865 * reconfigure. They might be read-only
1866 * or might be undergoing reshape etc.
e9dd1598
N
1867 * containername is e.g. md0, md_d1
1868 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1869 */
1870 return (*vers == '/' || *vers == '-');
3c558363
N
1871}
1872
e5408a32
DW
1873static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1874{
1875 return &ent->metadata_version[10+strlen(container)+1];
1876}
1877
83a379cf
LF
1878/**
1879 * signal_s() - Wrapper for sigaction() with signal()-like interface.
1880 * @sig: The signal to set the signal handler to.
1881 * @handler: The signal handler.
1882 *
1883 * Return: previous handler or SIG_ERR on failure.
1884 */
1885static inline sighandler_t signal_s(int sig, sighandler_t handler)
1886{
38cb95dd
MK
1887 struct sigaction new_act = {0};
1888 struct sigaction old_act = {0};
83a379cf
LF
1889
1890 new_act.sa_handler = handler;
83a379cf
LF
1891
1892 if (sigaction(sig, &new_act, &old_act) == 0)
1893 return old_act.sa_handler;
1894
1895 return SIG_ERR;
1896}
1897
2a24d7b6
DW
1898#ifdef DEBUG
1899#define dprintf(fmt, arg...) \
1ade5cc1
N
1900 fprintf(stderr, "%s: %s: "fmt, Name, __func__, ##arg)
1901#define dprintf_cont(fmt, arg...) \
2a24d7b6
DW
1902 fprintf(stderr, fmt, ##arg)
1903#else
1904#define dprintf(fmt, arg...) \
1ade5cc1
N
1905 ({ if (0) fprintf(stderr, "%s: %s: " fmt, Name, __func__, ##arg); 0; })
1906#define dprintf_cont(fmt, arg...) \
2a24d7b6
DW
1907 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1908#endif
7fcbfd7c 1909
1a0ee0ba
DK
1910static inline int xasprintf(char **strp, const char *fmt, ...) {
1911 va_list ap;
1912 int ret;
1913 va_start(ap, fmt);
089485cb 1914 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1915 va_end(ap);
1916 assert(ret >= 0);
1917 return ret;
1918}
2a24d7b6 1919
1ade5cc1
N
1920#ifdef DEBUG
1921#define pr_err(fmt, args...) fprintf(stderr, "%s: %s: "fmt, Name, __func__, ##args)
1922#else
d56dd607 1923#define pr_err(fmt, args...) fprintf(stderr, "%s: "fmt, Name, ##args)
1ade5cc1 1924#endif
e7b84f9d
N
1925#define cont_err(fmt ...) fprintf(stderr, " " fmt)
1926
9364dbfb
LG
1927#define pr_info(fmt, args...) printf("%s: "fmt, Name, ##args)
1928
503975b9
N
1929void *xmalloc(size_t len);
1930void *xrealloc(void *ptr, size_t len);
1931void *xcalloc(size_t num, size_t size);
1932char *xstrdup(const char *str);
1933
b5e64645
NB
1934#define LEVEL_MULTIPATH (-4)
1935#define LEVEL_LINEAR (-1)
1936#define LEVEL_FAULTY (-5)
1937
17f25ca6
NB
1938/* kernel module doesn't know about these */
1939#define LEVEL_CONTAINER (-100)
a322f70c 1940#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1941
56cb05c4
N
1942/* the kernel does know about this one ... */
1943#define LEVEL_NONE (-1000000)
1944
b5e64645
NB
1945/* faulty stuff */
1946
1947#define WriteTransient 0
1948#define ReadTransient 1
1949#define WritePersistent 2
1950#define ReadPersistent 3
1951#define WriteAll 4 /* doesn't go to device */
1952#define ReadFixable 5
1953#define Modes 6
1954
1955#define ClearErrors 31
1956#define ClearFaults 30
1957
1958#define AllPersist 100 /* internal use only */
1959#define NoPersist 101
1960
1961#define ModeMask 0x1f
1962#define ModeShift 5
fe6729fa 1963
fe6729fa
NB
1964#ifdef __TINYC__
1965#undef minor
1966#undef major
1967#undef makedev
1968#define minor(x) ((x)&0xff)
1969#define major(x) (((x)>>8)&0xff)
1970#define makedev(M,m) (((M)<<8) | (m))
1971#endif
1972
329dfc28
N
1973enum r0layout {
1974 RAID0_ORIG_LAYOUT = 1,
1975 RAID0_ALT_MULTIZONE_LAYOUT = 2,
1976};
1977
b640a252 1978/* for raid4/5/6 */
e86c9dd6
NB
1979#define ALGORITHM_LEFT_ASYMMETRIC 0
1980#define ALGORITHM_RIGHT_ASYMMETRIC 1
1981#define ALGORITHM_LEFT_SYMMETRIC 2
1982#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1983
1984/* Define non-rotating (raid4) algorithms. These allow
1985 * conversion of raid4 to raid5.
1986 */
1987#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1988#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1989
1990/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1991 * Firstly, the exact positioning of the parity block is slightly
1992 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1993 * of DDF.
1994 * Secondly, or order of datablocks over which the Q syndrome is computed
1995 * is different.
1996 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1997 * These layouts are from the DDFv1.2 spec.
1998 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1999 * leaves RLQ=3 as 'Vendor Specific'
2000 */
2001
2002#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
2003#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
2004#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
2005
b640a252
N
2006/* For every RAID5 algorithm we define a RAID6 algorithm
2007 * with exactly the same layout for data and parity, and
2008 * with the Q block always on the last device (N-1).
2009 * This allows trivial conversion from RAID5 to RAID6
2010 */
2011#define ALGORITHM_LEFT_ASYMMETRIC_6 16
2012#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
2013#define ALGORITHM_LEFT_SYMMETRIC_6 18
2014#define ALGORITHM_RIGHT_SYMMETRIC_6 19
2015#define ALGORITHM_PARITY_0_6 20
2016#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
2017
33a6535d
AW
2018/* Define PATH_MAX in case we don't use glibc or standard library does
2019 * not have PATH_MAX defined. Assume max path length is 4K characters.
2020 */
2021#ifndef PATH_MAX
2022#define PATH_MAX 4096
2023#endif
2024
9dad51d4
N
2025#define RESYNC_NONE -1
2026#define RESYNC_DELAYED -2
2027#define RESYNC_PENDING -3
1c294b5d
LZ
2028#define RESYNC_REMOTE -4
2029#define RESYNC_UNKNOWN -5
a0963a86 2030
480f3566
N
2031/* When using "GET_DISK_INFO" it isn't certain how high
2032 * we need to check. So we impose an absolute limit of
2033 * MAX_DISKS. This needs to be much more than the largest
2034 * number of devices any metadata can support. Currently
2035 * v1.x can support 1920
2036 */
2037#define MAX_DISKS 4096
2038
d04f65f4
N
2039/* Sometimes the 'size' value passed needs to mean "Maximum".
2040 * In those cases with use MAX_SIZE
2041 */
2042#define MAX_SIZE 1
822e393a
N
2043
2044/* We want to use unsigned numbers for sector counts, but need
2045 * a value for 'invalid'. Use '1'.
2046 */
2047#define INVALID_SECTORS 1
72ca9bcf
N
2048/* And another special number needed for --data_offset=variable */
2049#define VARIABLE_OFFSET 3
ced5fa8b 2050
6f2af6a4
KT
2051/**
2052 * is_container() - check if @level is &LEVEL_CONTAINER
2053 * @level: level value
2054 *
2055 * return:
2056 * 1 if level is equal to &LEVEL_CONTAINER, 0 otherwise.
2057 */
2058static inline int is_container(const int level)
2059{
2060 if (level == LEVEL_CONTAINER)
2061 return 1;
2062 return 0;
2063}
b823c8f9
MK
2064
2065#define STR_COMMON_NONE "none"
2066
2067/**
2068 * str_is_none() - check if @str starts with "none".
2069 * @str: string
2070 *
2071 * return:
2072 * true if string starts with "none", false otherwise.
2073 */
2074static inline bool str_is_none(char *str)
2075{
2076 if (strncmp(str, STR_COMMON_NONE, sizeof(STR_COMMON_NONE) - 1) == 0)
2077 return true;
2078 return false;
2079}