]> git.ipfire.org Git - thirdparty/mdadm.git/blame - mdadm.h
tests/23rdev-lifetime: fix a typo
[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;
51a9f2fc 436 struct dev_policy *pols;
fbfdcb06
AO
437};
438
67417d92
MT
439typedef enum mdadm_status {
440 MDADM_STATUS_SUCCESS = 0,
441 MDADM_STATUS_ERROR,
442 MDADM_STATUS_UNDEF,
443} mdadm_status_t;
444
e0d19036
NB
445enum mode {
446 ASSEMBLE=1,
447 BUILD,
448 CREATE,
449 MANAGE,
450 MISC,
451 MONITOR,
dd0781e5 452 GROW,
8382f19b 453 INCREMENTAL,
1f48664b 454 AUTODETECT,
5187a385 455 mode_count
e0d19036
NB
456};
457
64c4757e 458extern char short_options[];
6c9d9260 459extern char short_monitor_options[];
024768c4 460extern char short_bitmap_options[];
c06487ce 461extern char short_bitmap_auto_options[];
64c4757e 462extern struct option long_options[];
56eedc1a 463extern char Version[], Usage[], Help[], OptionHelp[],
5187a385 464 *mode_help[],
dd0781e5 465 Help_create[], Help_build[], Help_assemble[], Help_grow[],
8382f19b 466 Help_incr[],
e0d19036 467 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
64c4757e 468
997aed5d 469/* for option that don't have short equivilents, we assign arbitrary
1c7a808c 470 * numbers later than any 'short' character option.
997aed5d
NB
471 */
472enum special_options {
1c7a808c 473 AssumeClean = 300,
577fd104 474 WriteZeroes,
997aed5d
NB
475 BitmapChunk,
476 WriteBehind,
477 ReAdd,
478 NoDegraded,
479 Sparc22,
1c7a808c 480 BackupFile,
997aed5d 481 HomeHost,
589395d6 482 AutoHomeHost,
1f48664b 483 AutoDetect,
1770662b 484 Waitclean,
4cce4069 485 DetailPlatform,
33414a01 486 KillSubarray,
1c7a808c 487 UpdateSubarray,
edde9560 488 IncrementalPath,
1c7a808c
N
489 NoSharing,
490 HelpOptions,
491 Brief,
d11abe4b 492 NoDevices,
1c7a808c
N
493 ManageOpt,
494 Add,
f33a71f1 495 AddSpare,
01290056 496 AddJournal,
1c7a808c
N
497 Remove,
498 Fail,
70c55e36
N
499 Replace,
500 With,
1c7a808c
N
501 MiscOpt,
502 WaitOpt,
503 ConfigFile,
504 ChunkSize,
505 WriteMostly,
71574efb
N
506 FailFast,
507 NoFailFast,
1c7a808c
N
508 Layout,
509 Auto,
510 Force,
511 SuperMinor,
512 EMail,
513 ProgramOpt,
514 Increment,
515 Fork,
516 Bitmap,
517 RebuildMapOpt,
87f26d14 518 InvalidBackup,
20b60dcd 519 UdevRules,
b76b30e0 520 FreezeReshape,
2dddadb0 521 Continue,
08ca2adf 522 OffRootOpt,
c2ecf5f6 523 Prefer,
f7d3febc 524 KillOpt,
40c9a66a 525 DataOffset,
6d388a88 526 ExamineBB,
74db60b0
N
527 Dump,
528 Restore,
a740cf64 529 Action,
529e2aa5 530 Nodes,
7716570e 531 ClusterName,
4de90913 532 ClusterConfirm,
cc1799c3 533 WriteJournal,
5308f117 534 ConsistencyPolicy,
997aed5d
NB
535};
536
2568ce89
MK
537enum update_opt {
538 UOPT_NAME = 1,
539 UOPT_PPL,
540 UOPT_NO_PPL,
541 UOPT_BITMAP,
542 UOPT_NO_BITMAP,
543 UOPT_SUBARRAY_ONLY,
544 UOPT_SPARC22,
545 UOPT_SUPER_MINOR,
546 UOPT_SUMMARIES,
547 UOPT_RESYNC,
548 UOPT_UUID,
549 UOPT_HOMEHOST,
550 UOPT_HOME_CLUSTER,
551 UOPT_NODES,
552 UOPT_DEVICESIZE,
553 UOPT_BBL,
554 UOPT_NO_BBL,
555 UOPT_FORCE_NO_BBL,
556 UOPT_METADATA,
557 UOPT_REVERT_RESHAPE,
558 UOPT_LAYOUT_ORIGINAL,
559 UOPT_LAYOUT_ALTERNATE,
560 UOPT_LAYOUT_UNSPECIFIED,
561 UOPT_BYTEORDER,
562 UOPT_HELP,
2257de10
MK
563 UOPT_USER_ONLY,
564 /*
565 * Code specific options, cannot be set by the user
566 */
567 UOPT_SPEC_FORCE_ONE,
568 UOPT_SPEC_FORCE_ARRAY,
569 UOPT_SPEC_ASSEMBLE,
570 UOPT_SPEC_LINEAR_GROW_NEW,
571 UOPT_SPEC_LINEAR_GROW_UPDATE,
572 UOPT_SPEC__RESHAPE_PROGRESS,
573 UOPT_SPEC_WRITEMOSTLY,
574 UOPT_SPEC_READWRITE,
575 UOPT_SPEC_FAILFAST,
576 UOPT_SPEC_NOFAILFAST,
577 UOPT_SPEC_REVERT_RESHAPE_NOBACKUP,
2568ce89
MK
578 UOPT_UNDEFINED
579};
580extern void fprint_update_options(FILE *outf, enum update_opt update_mode);
581
f0ec6710
MN
582enum prefix_standard {
583 JEDEC,
584 IEC
585};
586
0aa2f15b
GJ
587enum bitmap_update {
588 NoUpdate,
589 NameUpdate,
7e6e839a 590 NodeNumUpdate,
0aa2f15b
GJ
591};
592
e22fe3ae
N
593enum flag_mode {
594 FlagDefault, FlagSet, FlagClear,
595};
596
0acda705
MK
597typedef enum {
598 ROLLBACK_METADATA_CHANGES,
599 APPLY_METADATA_CHANGES
600} change_dir_t;
601
64c4757e 602/* structures read from config file */
52826846
NB
603/* List of mddevice names and identifiers
604 * Identifiers can be:
605 * uuid=128-hex-uuid
606 * super-minor=decimal-minor-number-from-superblock
607 * devices=comma,separated,list,of,device,names,with,wildcards
608 *
609 * If multiple fields are present, the intersection of all matching
610 * devices is considered
611 */
98c6faba 612#define UnSet (0xfffe)
fa56eddb 613struct mddev_ident {
dd0781e5 614 char *devname;
aba69144 615
dd0781e5 616 int uuid_set;
3fa06e9d 617 int uuid[4];
67417d92 618 char name[MD_NAME_MAX + 1];
52826846 619
f21e18ca 620 int super_minor;
52826846 621
dd0781e5 622 char *devices; /* comma separated list of device
52826846
NB
623 * names with wild cards
624 */
dd0781e5 625 int level;
f21e18ca
N
626 int raid_disks;
627 int spare_disks;
82d9eba6 628 struct supertype *st;
dd0781e5
NB
629 int autof; /* 1 for normal, 2 for partitioned */
630 char *spare_group;
7ef02d01 631 char *bitmap_file;
c82f047c 632 int bitmap_fd;
dd0781e5 633
1771a6e2
N
634 char *container; /* /dev/whatever name of container, or
635 * uuid of container. You would expect
636 * this to be the 'devname' or UUID
dbb44303
N
637 * of some other entry.
638 */
639 char *member; /* subarray within a container */
640
fa56eddb 641 struct mddev_ident *next;
b179246f
N
642 union {
643 /* fields needed by different users of this structure */
644 int assembled; /* set when assembly succeeds */
645 };
fa56eddb 646};
64c4757e 647
9e33d556
N
648struct context {
649 int readonly;
650 int runstop;
651 int verbose;
9e33d556 652 int brief;
d11abe4b 653 int no_devices;
9e33d556
N
654 int force;
655 char *homehost;
656 int require_homehost;
657 char *prefer;
658 int export;
659 int test;
660 char *subarray;
3a87fa67 661 enum update_opt update;
9e33d556
N
662 int scan;
663 int SparcAdjust;
664 int autof;
665 int delay;
666 int freeze_reshape;
667 char *backup_file;
668 int invalid_backup;
a740cf64 669 char *action;
529e2aa5 670 int nodes;
7716570e 671 char *homecluster;
9e33d556
N
672};
673
e705e81b 674struct shape {
0acda705 675 char *dev;
e705e81b 676 int raiddisks;
0acda705 677 int delta_disks;
e705e81b 678 int sparedisks;
cc1799c3 679 int journaldisks;
e705e81b
N
680 int level;
681 int layout;
682 char *layout_str;
683 int chunk;
684 int bitmap_chunk;
685 char *bitmap_file;
686 int assume_clean;
577fd104 687 bool write_zeroes;
e705e81b
N
688 int write_behind;
689 unsigned long long size;
ae5dfc56 690 unsigned long long data_offset;
5308f117 691 int consistency_policy;
0acda705 692 change_dir_t direction;
e705e81b
N
693};
694
64c4757e 695/* List of device names - wildcards expanded */
a655e550 696struct mddev_dev {
64c4757e 697 char *devname;
c8e1a230
N
698 int disposition; /* 'a' for add, 'r' for remove, 'f' for fail,
699 * 'A' for re_add.
cd29a5c8
NB
700 * Not set for names read from .config
701 */
e22fe3ae
N
702 enum flag_mode writemostly;
703 enum flag_mode failfast;
70c55e36 704 int used; /* set when used */
72ca9bcf 705 long long data_offset;
a655e550
N
706 struct mddev_dev *next;
707};
64c4757e 708
682c7051
NB
709typedef struct mapping {
710 char *name;
711 int num;
712} mapping_t;
713
e0d19036 714struct mdstat_ent {
4dd2df09 715 char devnm[32];
e0d19036
NB
716 int active;
717 char *level;
7054da69 718 char *pattern; /* U for up, _ for down */
e0d19036 719 int percent; /* -1 if no resync */
f94c116f 720 int resync; /* 3 if check, 2 if reshape, 1 if resync, 0 if recovery */
549e9569
NB
721 int devcnt;
722 int raid_disks;
549e9569 723 char * metadata_version;
3b57c466
N
724 struct dev_member {
725 char *name;
726 struct dev_member *next;
1011e834 727 } *members;
e0d19036
NB
728 struct mdstat_ent *next;
729};
730
22a88995 731extern struct mdstat_ent *mdstat_read(int hold, int start);
a7a0d8a1 732extern void mdstat_close(void);
e0d19036 733extern void free_mdstat(struct mdstat_ent *ms);
cab9c67d 734extern int mdstat_wait(int seconds);
58a4ba2a 735extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
4dd2df09 736extern int mddev_busy(char *devnm);
3b57c466 737extern struct mdstat_ent *mdstat_by_component(char *name);
4dd2df09 738extern struct mdstat_ent *mdstat_by_subdev(char *subdev, char *container);
8382f19b
NB
739
740struct map_ent {
741 struct map_ent *next;
4dd2df09 742 char devnm[32];
1522c538 743 char metadata[20];
8382f19b 744 int uuid[4];
195254b8 745 int bad;
8382f19b
NB
746 char *path;
747};
4dd2df09 748extern int map_update(struct map_ent **mpp, char *devnm, char *metadata,
8382f19b 749 int uuid[4], char *path);
4dd2df09 750extern void map_remove(struct map_ent **map, char *devnm);
8382f19b 751extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
4dd2df09 752extern struct map_ent *map_by_devnm(struct map_ent **map, char *devnm);
eb2306f8 753extern void map_free(struct map_ent *map);
f2e55ecc 754extern struct map_ent *map_by_name(struct map_ent **map, char *name);
8382f19b
NB
755extern void map_read(struct map_ent **melp);
756extern int map_write(struct map_ent *mel);
4dd2df09 757extern void map_delete(struct map_ent **mapp, char *devnm);
8382f19b 758extern void map_add(struct map_ent **melp,
4dd2df09 759 char *devnm, char *metadata, int uuid[4], char *path);
ad5bc697
N
760extern int map_lock(struct map_ent **melp);
761extern void map_unlock(struct map_ent **melp);
cc700db3 762extern void map_fork(void);
e0d19036 763
e86c9dd6 764/* various details can be requested */
dab4a513
DW
765enum sysfs_read_flags {
766 GET_LEVEL = (1 << 0),
767 GET_LAYOUT = (1 << 1),
768 GET_COMPONENT = (1 << 2),
769 GET_CHUNK = (1 << 3),
770 GET_CACHE = (1 << 4),
771 GET_MISMATCH = (1 << 5),
772 GET_VERSION = (1 << 6),
773 GET_DISKS = (1 << 7),
dab4a513 774 GET_SAFEMODE = (1 << 9),
c0c1acd6
N
775 GET_BITMAP_LOCATION = (1 << 10),
776
777 GET_DEVS = (1 << 20), /* gets role, major, minor */
778 GET_OFFSET = (1 << 21),
779 GET_SIZE = (1 << 22),
780 GET_STATE = (1 << 23),
781 GET_ERROR = (1 << 24),
5aa644c6 782 GET_ARRAY_STATE = (1 << 25),
5308f117 783 GET_CONSISTENCY_POLICY = (1 << 26),
ae7d61e3 784 GET_DEVS_ALL = (1 << 27),
dab4a513 785};
e86c9dd6 786
90fd7001
MK
787#define SYSFS_MAX_BUF_SIZE 64
788
14a86579
MT
789extern void sysfs_get_container_devnm(struct mdinfo *mdi, char *buf);
790
e86c9dd6 791/* If fd >= 0, get the array it is open on,
4dd2df09 792 * else use devnm.
e86c9dd6 793 */
4dd2df09 794extern int sysfs_open(char *devnm, char *devname, char *attr);
dae13137 795extern int sysfs_init(struct mdinfo *mdi, int fd, char *devnm);
a37563c9 796extern void sysfs_init_dev(struct mdinfo *mdi, dev_t devid);
7e0f6979 797extern void sysfs_free(struct mdinfo *sra);
4dd2df09 798extern struct mdinfo *sysfs_read(int fd, char *devnm, unsigned long options);
1770662b
DW
799extern int sysfs_attr_match(const char *attr, const char *str);
800extern int sysfs_match_word(const char *word, char **list);
7e0f6979 801extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 802 char *name, char *val);
7e0f6979 803extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 804 char *name, unsigned long long val);
012a8641
JS
805extern int sysfs_set_num_signed(struct mdinfo *sra, struct mdinfo *dev,
806 char *name, long long val);
97590376 807extern int sysfs_uevent(struct mdinfo *sra, char *event);
7236ee7a
N
808extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
809 char *name);
810extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
7e0f6979 811extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
e86c9dd6 812 char *name, unsigned long long *val);
2eba8496
N
813extern int sysfs_fd_get_two(int fd, unsigned long long *v1, unsigned long long *v2);
814extern int sysfs_get_two(struct mdinfo *sra, struct mdinfo *dev,
815 char *name, unsigned long long *v1, unsigned long long *v2);
7236ee7a 816extern int sysfs_fd_get_str(int fd, char *val, int size);
bc77ed53
DW
817extern int sysfs_attribute_available(struct mdinfo *sra, struct mdinfo *dev,
818 char *name);
93ecfa01
N
819extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
820 char *name, char *val, int size);
8ed3e5e1 821extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
de23e12a 822extern int sysfs_set_array(struct mdinfo *info);
2904b26f 823extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd, int resume);
f1665f72 824extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
4dd2df09 825extern int sysfs_unique_holder(char *devnm, long rdev);
bc77ed53 826extern int sysfs_freeze_array(struct mdinfo *sra);
efc67e8e 827extern int sysfs_wait(int fd, int *msec);
193b6c0b 828extern int load_sys(char *path, char *buf, int len);
b2514242 829extern int zero_disk_range(int fd, unsigned long long sector, size_t count);
130994cb
AK
830extern int reshape_prepare_fdlist(char *devname,
831 struct mdinfo *sra,
832 int raid_disks,
833 int nrdisks,
834 unsigned long blocks,
835 char *backup_file,
836 int *fdlist,
837 unsigned long long *offsets);
838extern void reshape_free_fdlist(int *fdlist,
839 unsigned long long *offsets,
840 int size);
e6e9d47b
AK
841extern int reshape_open_backup_file(char *backup,
842 int fd,
843 char *devname,
844 long blocks,
845 int *fdlist,
a93f87ee 846 unsigned long long *offsets,
54ded86f 847 char *sysfs_name,
a93f87ee 848 int restart);
1c009fc2
AK
849extern unsigned long compute_backup_blocks(int nchunk, int ochunk,
850 unsigned int ndata, unsigned int odata);
54ded86f
N
851extern char *locate_backup(char *name);
852extern char *make_backup(char *name);
e86c9dd6
NB
853
854extern int save_stripes(int *source, unsigned long long *offsets,
855 int raid_disks, int chunk_size, int level, int layout,
856 int nwrites, int *dest,
a6288483
N
857 unsigned long long start, unsigned long long length,
858 char *buf);
353632d9
NB
859extern int restore_stripes(int *dest, unsigned long long *offsets,
860 int raid_disks, int chunk_size, int level, int layout,
861 int source, unsigned long long read_offset,
2fcb75ae
AK
862 unsigned long long start, unsigned long long length,
863 char *src_buf);
df38df30 864extern bool sysfs_is_libata_allow_tpm_enabled(const int verbose);
e86c9dd6 865
52826846
NB
866#ifndef Sendmail
867#define Sendmail "/usr/lib/sendmail -t"
868#endif
869
773135f5 870#define SYSLOG_FACILITY LOG_DAEMON
5f21d674 871extern char *map_num_s(mapping_t *map, int num);
682c7051
NB
872extern char *map_num(mapping_t *map, int num);
873extern int map_name(mapping_t *map, char *name);
329dfc28
N
874extern mapping_t r0layout[], r5layout[], r6layout[],
875 pers[], modes[], faultylayout[];
2568ce89 876extern mapping_t consistency_policies[], sysfs_array_states[], update_options[];
682c7051 877
c2ecf5f6
N
878extern char *map_dev_preferred(int major, int minor, int create,
879 char *prefer);
880static inline char *map_dev(int major, int minor, int create)
881{
882 return map_dev_preferred(major, minor, create, NULL);
883}
64c4757e 884
4389ce73
MT
885/**
886 * is_fd_valid() - check file descriptor.
887 * @fd: file descriptor.
888 *
889 * The function checks if @fd is nonnegative integer and shall be used only
890 * to verify open() result.
891 */
892static inline int is_fd_valid(int fd)
893{
894 return (fd > -1);
895}
896
70f1ff42
LF
897/**
898 * is_level456() - check whether given level is between inclusive 4 and 6.
899 * @level: level to check.
900 *
901 * Return: true if condition is met, false otherwise
902 */
903static inline bool is_level456(int level)
904{
905 return (level >= 4 && level <= 6);
906}
907
4389ce73
MT
908/**
909 * close_fd() - verify, close and unset file descriptor.
910 * @fd: pointer to file descriptor.
911 *
912 * The function closes and invalidates file descriptor if appropriative. It
913 * ignores incorrect file descriptor quitely to simplify error handling.
914 */
915static inline void close_fd(int *fd)
916{
917 if (is_fd_valid(*fd) && close(*fd) == 0)
918 *fd = -1;
919}
920
549e9569 921struct active_array;
2e735d19 922struct metadata_update;
64c4757e 923
80bf9135 924/* 'struct reshape' records the intermediate states of
999b4972
N
925 * a general reshape.
926 * The starting geometry is converted to the 'before' geometry
927 * by at most an atomic level change. They could be the same.
928 * Similarly the 'after' geometry is converted to the final
929 * geometry by at most a level change.
930 * Note that 'before' and 'after' must have the same level.
931 * 'blocks' is the minimum number of sectors for a reshape unit.
932 * This will be a multiple of the stripe size in each of the
933 * 'before' and 'after' geometries.
934 * If 'blocks' is 0, no restriping is necessary.
89ecd3cf
N
935 * 'min_offset_change' is the minimum change to data_offset to
936 * allow the reshape to happen. It is at least the larger of
937 * the old and new chunk sizes, and typically the same as 'blocks'
938 * divided by number of data disks.
999b4972
N
939 */
940struct reshape {
941 int level;
942 int parity; /* number of parity blocks/devices */
943 struct {
944 int layout;
945 int data_disks;
946 } before, after;
947 unsigned long long backup_blocks;
89ecd3cf 948 unsigned long long min_offset_change;
999b4972
N
949 unsigned long long stripes; /* number of old stripes that comprise 'blocks'*/
950 unsigned long long new_size; /* New size of array in sectors */
951};
952
1fef0c6f
MT
953/**
954 * struct dev_policy - Data structure for policy management.
955 * @next: pointer to next dev_policy.
956 * @name: policy name, category.
957 * @metadata: the metadata type it affects.
958 * @value: value of the policy.
959 *
960 * The functions to manipulate dev_policy lists do not free elements, so they must be statically
961 * allocated. @name and @metadata can be compared by address.
962 */
963typedef struct dev_policy {
964 struct dev_policy *next;
965 char *name;
966 const char *metadata;
967 const char *value;
968} dev_policy_t;
969
7054da69 970/* A superswitch provides entry point to a metadata handler.
6adfd3af 971 *
74db60b0 972 * The superswitch primarily operates on some "metadata" that
6adfd3af
NB
973 * is accessed via the 'supertype'.
974 * This metadata has one of three possible sources.
975 * 1/ It is read from a single device. In this case it may not completely
976 * describe the array or arrays as some information might be on other
977 * devices.
978 * 2/ It is read from all devices in a container. In this case all
979 * information is present.
980 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
981 * be complete once enough ->add_to_super calls have completed.
982 *
983 * When creating an array inside a container, the metadata will be
984 * formed by a combination of 2 and 3. The metadata or the array is read,
985 * then new information is added.
986 *
987 * The metadata must sometimes have a concept of a 'current' array
988 * and a 'current' device.
989 * The 'current' array is set by init_super to be the newly created array,
990 * or is set by super_by_fd when it finds it is looking at an array inside
991 * a container.
992 *
993 * The 'current' device is either the device that the metadata was read from
994 * in case 1, or the last device added by add_to_super in case 3.
995 * Case 2 does not identify a 'current' device.
996 */
f9ce90ba 997extern struct superswitch {
6adfd3af
NB
998
999 /* Used to report details of metadata read from a component
1000 * device. ->load_super has been called.
1001 */
3da92f27 1002 void (*examine_super)(struct supertype *st, char *homehost);
061f2c6a 1003 void (*brief_examine_super)(struct supertype *st, int verbose);
4737ae25 1004 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
0d726f17 1005 void (*export_examine_super)(struct supertype *st);
6d388a88 1006 int (*examine_badblocks)(struct supertype *st, int fd, char *devname);
74db60b0 1007 int (*copy_metadata)(struct supertype *st, int from, int to);
6adfd3af
NB
1008
1009 /* Used to report details of an active array.
1010 * ->load_super was possibly given a 'component' string.
1011 */
b771faef
BK
1012 void (*detail_super)(struct supertype *st, char *homehost,
1013 char *subarray);
1014 void (*brief_detail_super)(struct supertype *st, char *subarray);
0d726f17 1015 void (*export_detail_super)(struct supertype *st);
6adfd3af 1016
4cce4069 1017 /* Optional: platform hardware / firmware details */
9eafa1de
MN
1018 int (*detail_platform)(int verbose, int enumerate_only, char *controller_path);
1019 int (*export_detail_platform)(int verbose, char *controller_path);
4cce4069 1020
6adfd3af
NB
1021 /* Used:
1022 * to get uuid to storing in bitmap metadata
1023 * and 'reshape' backup-data metadata
1024 * To see if a device is being re-added to an array it was part of.
1025 */
3da92f27 1026 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
6adfd3af 1027
8592f29d 1028 /* Extract generic details from metadata. This could be details about
6adfd3af
NB
1029 * the container, or about an individual array within the container.
1030 * The determination is made either by:
1031 * load_super being given a 'component' string.
1032 * validate_geometry determining what to create.
d2ca6449
NB
1033 * The info includes both array information and device information.
1034 * The particular device should be:
1035 * The last device added by add_to_super
1036 * The device the metadata was loaded from by load_super
a5d85af7
N
1037 * If 'map' is present, then it is an array raid_disks long
1038 * (raid_disk must already be set and correct) and it is filled
1039 * with 1 for slots that are thought to be active and 0 for slots which
1040 * appear to be failed/missing.
95eeceeb 1041 * *info is zeroed out before data is added.
6adfd3af 1042 */
a5d85af7 1043 void (*getinfo_super)(struct supertype *st, struct mdinfo *info, char *map);
5c4cd5da 1044 struct mdinfo *(*getinfo_super_disks)(struct supertype *st);
6adfd3af 1045 /* Check if the given metadata is flagged as belonging to "this"
9362c1c8 1046 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
6adfd3af 1047 */
3da92f27 1048 int (*match_home)(struct supertype *st, char *homehost);
6adfd3af
NB
1049
1050 /* Make one of several generic modifications to metadata
1051 * prior to assembly (or other times).
1052 * sparc2.2 - first bug in early 0.90 metadata
1053 * super-minor - change name of 0.90 metadata
1054 * summaries - 'correct' any redundant data
1055 * resync - mark array as dirty to trigger a resync.
1056 * uuid - set new uuid - only 0.90 or 1.x
1057 * name - change the name of the array (where supported)
1058 * homehost - change which host this array is tied to.
1059 * devicesize - If metadata is at start of device, change recorded
1060 * device size to match actual device size
1061 * byteorder - swap bytes for 0.90 metadata
1062 *
1063 * force-one - mark that device as uptodate, not old or failed.
1064 * force-array - mark array as clean if it would not otherwise
1065 * assemble
1066 * assemble - not sure how this is different from force-one...
1067 * linear-grow-new - add a new device to a linear array, but don't
1068 * change the size: so superblock still matches
1069 * linear-grow-update - now change the size of the array.
16715c01
DL
1070 * writemostly - set the WriteMostly1 bit in the superblock devflags
1071 * readwrite - clear the WriteMostly1 bit in the superblock devflags
71574efb
N
1072 * failfast - set the FailFast1 bit in the superblock
1073 * nofailfast - clear the FailFast1 bit
199f1a1f
N
1074 * no-bitmap - clear any record that a bitmap is present.
1075 * bbl - add a bad-block-log if possible
6dd16dac
N
1076 * no-bbl - remove any bad-block-log is it is empty.
1077 * force-no-bbl - remove any bad-block-log even if empty.
199f1a1f
N
1078 * revert-reshape - If a reshape is in progress, modify metadata so
1079 * it will resume going in the opposite direction.
6adfd3af 1080 */
68c7d6d7 1081 int (*update_super)(struct supertype *st, struct mdinfo *info,
03312b52 1082 enum update_opt update,
e5eac01f
NB
1083 char *devname, int verbose,
1084 int uuid_set, char *homehost);
6adfd3af
NB
1085
1086 /* Create new metadata for new array as described. This could
1087 * be a new container, or an array in a pre-existing container.
1088 * Also used to zero metadata prior to writing it to invalidate old
1089 * metadata.
1090 */
3da92f27 1091 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
5308f117 1092 struct shape *s, char *name,
83cd1e97
N
1093 char *homehost, int *uuid,
1094 unsigned long long data_offset);
6adfd3af
NB
1095
1096 /* update the metadata to include new device, either at create or
1097 * when hot-adding a spare.
1098 */
f20c3968 1099 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
72ca9bcf
N
1100 int fd, char *devname,
1101 unsigned long long data_offset);
1a64be56
LM
1102 /* update the metadata to delete a device,
1103 * when hot-removing.
1104 */
1105 int (*remove_from_super)(struct supertype *st, mdu_disk_info_t *dinfo);
6adfd3af
NB
1106
1107 /* Write metadata to one device when fixing problems or adding
1108 * a new device.
1109 */
3da92f27 1110 int (*store_super)(struct supertype *st, int fd);
6adfd3af
NB
1111
1112 /* Write all metadata for this array.
1113 */
111d01fc 1114 int (*write_init_super)(struct supertype *st);
e2f408a4
N
1115 /* Check if metadata read from one device is compatible with an array,
1116 * used when assembling an array, or pseudo-assembling was with
1117 * "--examine --brief"
1118 * If "st" has not yet been loaded the superblock from, "tst" is
1119 * moved in, otherwise the superblock in 'st' is compared with
1120 * 'tst'.
1121 */
c7b8547c
MT
1122 int (*compare_super)(struct supertype *st, struct supertype *tst,
1123 int verbose);
e2f408a4
N
1124 /* Load metadata from a single device. If 'devname' is not NULL
1125 * print error messages as appropriate */
3da92f27 1126 int (*load_super)(struct supertype *st, int fd, char *devname);
e2f408a4
N
1127 /* 'fd' is a 'container' md array - load array metadata from the
1128 * whole container.
1129 */
2b959fbf 1130 int (*load_container)(struct supertype *st, int fd, char *devname);
e2f408a4
N
1131 /* If 'arg' is a valid name of this metadata type, allocate and
1132 * return a 'supertype' for the particular minor version */
82d9eba6 1133 struct supertype * (*match_metadata_desc)(char *arg);
e2f408a4
N
1134 /* If a device has the given size, and the data_offset has been
1135 * requested - work out how much space is available for data.
1136 * This involves adjusting for reserved space (e.g. bitmaps)
1137 * and for any rounding.
1138 * 'mdadm' only calls this for existing arrays where a possible
1139 * spare is being added. However some super-handlers call it
1140 * internally from validate_geometry when creating an array.
1141 */
387fcd59
N
1142 __u64 (*avail_size)(struct supertype *st, __u64 size,
1143 unsigned long long data_offset);
fbfdcb06
AO
1144 /*
1145 * Return spare criteria for array:
1146 * - minimum disk size can be used in array;
4b57ecf6 1147 * - sector size can be used in array.
e2f408a4 1148 */
f6562011
MT
1149 mdadm_status_t (*get_spare_criteria)(struct supertype *st, char *mddev_path,
1150 struct spare_criteria *sc);
e2f408a4
N
1151 /* Find somewhere to put a bitmap - possibly auto-size it - and
1152 * update the metadata to record this. The array may be newly
1153 * created, in which case data_size may be updated, or it might
1154 * already exist. Metadata handler can know if init_super
1155 * has been called, but not write_init_super.
2ec2b7e9
JS
1156 * 0: Success
1157 * -Exxxx: On error
e2f408a4 1158 */
3da92f27 1159 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
199171a2 1160 int delay, int write_behind,
f9c25f1d 1161 unsigned long long size, int may_change, int major);
b554ab5c
JR
1162 /* Perform additional setup required to activate a bitmap.
1163 */
1164 int (*set_bitmap)(struct supertype *st, struct mdinfo *info);
e2f408a4
N
1165 /* Seek 'fd' to start of write-intent-bitmap. Must be an
1166 * md-native format bitmap
1167 */
b138214f 1168 int (*locate_bitmap)(struct supertype *st, int fd, int node_num);
e2f408a4
N
1169 /* if add_internal_bitmap succeeded for existing array, this
1170 * writes it out.
1171 */
0aa2f15b 1172 int (*write_bitmap)(struct supertype *st, int fd, enum bitmap_update update);
e2f408a4 1173 /* Free the superblock and any other allocated data */
3da92f27 1174 void (*free_super)(struct supertype *st);
78e44928
NB
1175
1176 /* validate_geometry is called with an st returned by
1177 * match_metadata_desc.
e2f408a4 1178 * It should check that the geometry described is compatible with
78e44928
NB
1179 * the metadata type. It will be called repeatedly as devices
1180 * added to validate changing size and new devices. If there are
1181 * inter-device dependencies, it should record sufficient details
1182 * so these can be validated.
30f58b22 1183 * Both 'size' and '*freesize' are in sectors. chunk is KiB.
ecbd9e81
N
1184 * Return value is:
1185 * 1: everything is OK
1186 * 0: not OK for some reason - if 'verbose', then error was reported.
1187 * -1: st->sb was NULL, 'subdev' is a member of a container of this
e2f408a4 1188 * type, but array is not acceptable for some reason
ecbd9e81 1189 * message was reported even if verbose is 0.
78e44928 1190 */
17f25ca6
NB
1191 int (*validate_geometry)(struct supertype *st, int level, int layout,
1192 int raiddisks,
c21e737b 1193 int *chunk, unsigned long long size,
af4348dd 1194 unsigned long long data_offset,
2c514b71 1195 char *subdev, unsigned long long *freesize,
5308f117 1196 int consistency_policy, int verbose);
598f0d58 1197
1fef0c6f
MT
1198 /**
1199 * test_and_add_drive_policies() - test new and add custom policies from metadata handler.
1200 * @pols: list of currently recorded policies.
1201 * @disk_fd: file descriptor of the device to check.
1202 * @verbose: verbose flag.
1203 *
1204 * Used by IMSM to verify all drives in container/array, against requirements not recored
1205 * in superblock, like controller type for IMSM. It should check all drives even if
1206 * they are not actually used, because mdmon or kernel are free to use any drive assigned to
1207 * container automatically.
1208 *
1209 * Generating and comparison methods belong to metadata handler. It is not mandatory to be
1210 * implemented.
1211 *
1212 * Return: MDADM_STATUS_SUCCESS is expected on success.
1213 */
1214 mdadm_status_t (*test_and_add_drive_policies)(dev_policy_t **pols, int disk_fd,
1215 const int verbose);
1216
e2f408a4
N
1217 /* Return a linked list of 'mdinfo' structures for all arrays
1218 * in the container. For non-containers, it is like
1219 * getinfo_super with an allocated mdinfo.*/
00bbdbda 1220 struct mdinfo *(*container_content)(struct supertype *st, char *subarray);
30f58b22
DW
1221 /* query the supertype for default geometry */
1222 void (*default_geometry)(struct supertype *st, int *level, int *layout, int *chunk); /* optional */
33414a01 1223 /* Permit subarray's to be deleted from inactive containers */
3364781b
BK
1224 int (*kill_subarray)(struct supertype *st,
1225 char *subarray_id); /* optional */
03312b52
MK
1226 /**
1227 * update_subarray() - Permit subarray to be modified.
1228 * @st: Supertype.
1229 * @subarray: Subarray name.
1230 * @update: Update option.
1231 * @ident: Optional identifiers.
1232 */
a951a4f7 1233 int (*update_subarray)(struct supertype *st, char *subarray,
03312b52 1234 enum update_opt update, struct mddev_ident *ident);
7bc71196
DW
1235 /* Check if reshape is supported for this external format.
1236 * st is obtained from super_by_fd() where st->subarray[0] is
1237 * initialized to indicate if reshape is being performed at the
1238 * container or subarray level
1239 */
0acda705
MK
1240
1241 int (*reshape_super)(struct supertype *st, struct shape *shape, struct context *c);
999b4972
N
1242 int (*manage_reshape)( /* optional */
1243 int afd, struct mdinfo *sra, struct reshape *reshape,
1244 struct supertype *st, unsigned long blocks,
1245 int *fds, unsigned long long *offsets,
1246 int dests, int *destfd, unsigned long long *destoffsets);
598f0d58 1247
549e9569 1248/* for mdmon */
cba0191b 1249 int (*open_new)(struct supertype *c, struct active_array *a,
60815698 1250 int inst);
ed9d66aa
NB
1251
1252 /* Tell the metadata handler the current state of the array.
1253 * This covers whether it is known to be consistent (no pending writes)
01f157d7 1254 * and how far along a resync is known to have progressed
ed9d66aa
NB
1255 * (in a->resync_start).
1256 * resync status is really irrelevant if the array is not consistent,
1257 * but some metadata (DDF!) have a place to record the distinction.
1011e834 1258 * If 'consistent' is '2', then the array can mark it dirty if a
01f157d7
N
1259 * resync/recovery/whatever is required, or leave it clean if not.
1260 * Return value is 0 dirty (not consistent) and 1 if clean.
1261 * it is only really important if consistent is passed in as '2'.
4e5528c6 1262 */
01f157d7 1263 int (*set_array_state)(struct active_array *a, int consistent);
7a7cc504
NB
1264
1265 /* When the state of a device might have changed, we call set_disk to
1266 * tell the metadata what the current state is.
1267 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
1268 * transitions.
1269 * set_disk might be called when the state of the particular disk has
1270 * not in fact changed.
1271 */
8d45d196 1272 void (*set_disk)(struct active_array *a, int n, int state);
2e735d19
NB
1273 void (*sync_metadata)(struct supertype *st);
1274 void (*process_update)(struct supertype *st,
1275 struct metadata_update *update);
5fe6f031
N
1276 /* Prepare updates allocates extra memory that might be
1277 * needed. If the update cannot be understood, return 0.
1278 */
1279 int (*prepare_update)(struct supertype *st,
edd8d13c 1280 struct metadata_update *update);
549e9569 1281
6c3fb95c
NB
1282 /* activate_spare will check if the array is degraded and, if it
1283 * is, try to find some spare space in the container.
1284 * On success, it add appropriate updates (For process_update) to
1285 * to the 'updates' list and returns a list of 'mdinfo' identifying
1286 * the device, or devices as there might be multiple missing
1287 * devices and multiple spares available.
1288 */
1289 struct mdinfo *(*activate_spare)(struct active_array *a,
1290 struct metadata_update **updates);
549e9569 1291
ba53ea59
AK
1292 /* for external backup area */
1293 int (*recover_backup)(struct supertype *st, struct mdinfo *info);
1294
0c21b485
PB
1295 /* validate container after assemble */
1296 int (*validate_container)(struct mdinfo *info);
1297
5308f117
AP
1298 /* write initial empty PPL on device */
1299 int (*write_init_ppl)(struct supertype *st, struct mdinfo *info, int fd);
1300
2432ce9b
AP
1301 /* validate ppl before assemble */
1302 int (*validate_ppl)(struct supertype *st, struct mdinfo *info,
1303 struct mdinfo *disk);
1304
1ab97c97
TM
1305 /* records new bad block in metadata */
1306 int (*record_bad_block)(struct active_array *a, int n,
1307 unsigned long long sector, int length);
1308
42d902d9
TM
1309 /* clears bad block from metadata */
1310 int (*clear_bad_block)(struct active_array *a, int n,
1311 unsigned long long sector, int length);
1312
1313 /* get list of bad blocks from metadata */
1314 struct md_bb *(*get_bad_blocks)(struct active_array *a, int n);
1315
f277ce36 1316 int swapuuid; /* true if uuid is bigending rather than hostendian */
d03373f1 1317 int external;
4cce4069 1318 const char *name; /* canonical metadata name */
0f22b998 1319} *superlist[];
82d9eba6 1320
0f22b998
N
1321extern struct superswitch super0, super1;
1322extern struct superswitch super_imsm, super_ddf;
0592faeb 1323extern struct superswitch mbr, gpt;
cdddbdbc 1324
420dafcd
N
1325void imsm_set_no_platform(int v);
1326
edd8d13c
NB
1327struct metadata_update {
1328 int len;
1329 char *buf;
1330 void *space; /* allocated space that monitor will use */
cb23f1f4
N
1331 void **space_list; /* list of allocated spaces that monitor can
1332 * use or that it returned.
1333 */
edd8d13c
NB
1334 struct metadata_update *next;
1335};
1336
6adfd3af
NB
1337/* A supertype holds a particular collection of metadata.
1338 * It identifies the metadata type by the superswitch, and the particular
1339 * sub-version of that metadata type.
1340 * metadata read in or created is stored in 'sb' and 'info'.
1341 * There are also fields used by mdmon to track containers.
1342 *
352452c3
N
1343 * A supertype may refer to:
1344 * Just an array, possibly in a container
1345 * A container, not identifying any particular array
1346 * Info read from just one device, not yet fully describing the array/container.
1347 *
1348 *
6adfd3af
NB
1349 * A supertype is created by:
1350 * super_by_fd
1351 * guess_super
1352 * dup_super
1353 */
82d9eba6
NB
1354struct supertype {
1355 struct superswitch *ss;
1356 int minor_version;
ea329559 1357 int max_devs;
4dd2df09 1358 char container_devnm[32]; /* devnm of container */
64557c33 1359 void *sb;
111d01fc 1360 void *info;
afa368f4
N
1361 void *other; /* Hack used to convert v0.90 to v1.0 */
1362 unsigned long long devsize;
23bf42cc 1363 unsigned long long data_offset; /* used by v1.x only */
df3346e6
LM
1364 int ignore_hw_compat; /* used to inform metadata handlers that it should ignore
1365 HW/firmware related incompatability to load metadata.
1366 Used when examining metadata to display content of disk
1367 when user has no hw/firmare compatible system.
1368 */
edd8d13c
NB
1369 struct metadata_update *updates;
1370 struct metadata_update **update_tail;
1371
549e9569
NB
1372 /* extra stuff used by mdmon */
1373 struct active_array *arrays;
549e9569 1374 int sock; /* listen to external programs */
4dd2df09
N
1375 char devnm[32]; /* e.g. md0. This appears in metadata_version:
1376 * external:/md0/12
1377 */
549e9569 1378 int devcnt;
68226a80 1379 int retry_soon;
529e2aa5 1380 int nodes;
7716570e 1381 char *cluster_name;
549e9569
NB
1382
1383 struct mdinfo *devs;
1384
82d9eba6 1385};
f9ce90ba 1386
4725bc31 1387extern struct supertype *super_by_fd(int fd, char **subarray);
54887ad8
N
1388enum guess_types { guess_any, guess_array, guess_partitions };
1389extern struct supertype *guess_super_type(int fd, enum guess_types guess_type);
1390static inline struct supertype *guess_super(int fd) {
1391 return guess_super_type(fd, guess_any);
1392}
3da92f27 1393extern struct supertype *dup_super(struct supertype *st);
beae1dfe 1394extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
32971509 1395extern int get_dev_sector_size(int fd, char *dname, unsigned int *sectsizep);
3a371610 1396extern int must_be_container(int fd);
a7c6e3fb 1397void wait_for(char *dev, int fd);
f9ce90ba 1398
4e8d9f0a 1399extern char pol_act[], pol_domain[], pol_metadata[], pol_auto[];
5527fc74
N
1400
1401/* iterate over the sublist starting at list, having the same
1402 * 'name' as 'list', and matching the given metadata (Where
1403 * NULL matches anything
1404 */
1405#define pol_for_each(item, list, _metadata) \
1406 for (item = list; \
1407 item && item->name == list->name; \
1408 item = item->next) \
1409 if (!(!_metadata || !item->metadata || _metadata == item->metadata)) \
1410 ; else
1411
1412/*
1413 * policy records read from mdadm are largely just name-value pairs.
1414 * The names are constants, not strdupped
1415 */
1416struct pol_rule {
1417 struct pol_rule *next;
1418 char *type; /* rule_policy or rule_part */
1419 struct rule {
1420 struct rule *next;
1421 char *name;
1422 char *value;
1423 char *dups; /* duplicates of 'value' with a partNN appended */
1424 } *rule;
1425};
1426
1427extern char rule_policy[], rule_part[];
1428extern char rule_path[], rule_type[];
1429extern char type_part[], type_disk[];
1430
1431extern void policyline(char *line, char *type);
4e8d9f0a 1432extern void policy_add(char *type, ...);
5527fc74
N
1433extern void policy_free(void);
1434
cd72f9d1 1435extern struct dev_policy *path_policy(char **paths, char *type);
5527fc74 1436extern struct dev_policy *disk_policy(struct mdinfo *disk);
4dd2df09 1437extern struct dev_policy *devid_policy(int devid);
5527fc74
N
1438extern void dev_policy_free(struct dev_policy *p);
1439
e78dda3b 1440extern void pol_add(struct dev_policy **pol, char *name, char *val, char *metadata);
5527fc74
N
1441extern struct dev_policy *pol_find(struct dev_policy *pol, char *name);
1442
5a2e194c
MT
1443extern mdadm_status_t drive_test_and_add_policies(struct supertype *st, dev_policy_t **pols,
1444 int fd, const int verbose);
1445extern mdadm_status_t sysfs_test_and_add_drive_policies(struct supertype *st, dev_policy_t **pols,
1446 struct mdinfo *mdi, const int verbose);
1447extern mdadm_status_t mddev_test_and_add_drive_policies(struct supertype *st, dev_policy_t **pols,
1448 int array_fd, const int verbose);
1449
e3bb5f14
N
1450enum policy_action {
1451 act_default,
1452 act_include,
1453 act_re_add,
d2db3045
N
1454 act_spare, /* This only applies to bare devices */
1455 act_spare_same_slot, /* this allows non-bare devices,
1456 * but only if recent removal */
1457 act_force_spare, /* this allow non-bare devices in any case */
e3bb5f14
N
1458 act_err
1459};
1460
1461extern int policy_action_allows(struct dev_policy *plist, const char *metadata,
1462 enum policy_action want);
1463extern int disk_action_allows(struct mdinfo *disk, const char *metadata,
1464 enum policy_action want);
f5f12c84
N
1465
1466struct domainlist {
1467 struct domainlist *next;
2cda7640 1468 const char *dom;
f5f12c84
N
1469};
1470
1471extern int domain_test(struct domainlist *dom, struct dev_policy *pol,
1472 const char *metadata);
1473extern struct domainlist *domain_from_array(struct mdinfo *mdi,
1474 const char *metadata);
4dd2df09 1475extern void domainlist_add_dev(struct domainlist **dom, int devid,
e78dda3b 1476 const char *metadata);
f5f12c84
N
1477extern void domain_free(struct domainlist *dl);
1478extern void domain_merge(struct domainlist **domp, struct dev_policy *pol,
1479 const char *metadata);
e78dda3b 1480void domain_add(struct domainlist **domp, char *domain);
f5f12c84 1481
403410eb
PC
1482extern void policy_save_path(char *id_path, struct map_ent *array);
1483extern int policy_check_path(struct mdinfo *disk, struct map_ent *array);
1484
b0681598
MD
1485extern void sysfs_rules_apply(char *devnm, struct mdinfo *dev);
1486extern void sysfsline(char *line);
1487
838acbc2
NB
1488#if __GNUC__ < 3
1489struct stat64;
1490#endif
1491
45e878bb
NB
1492#define HAVE_NFTW we assume
1493#define HAVE_FTW
1494
2df1f269 1495#ifdef __UCLIBC__
45e878bb 1496# include <features.h>
2df1f269
BRF
1497# ifndef __UCLIBC_HAS_LFS__
1498# define lseek64 lseek
1499# endif
45e878bb
NB
1500# ifndef __UCLIBC_HAS_FTW__
1501# undef HAVE_FTW
1502# undef HAVE_NFTW
1503# endif
1504#endif
1505
1506#ifdef __dietlibc__
1507# undef HAVE_NFTW
1508#endif
1509
e39b673e 1510#if defined(__KLIBC__)
1511# undef HAVE_NFTW
1512# undef HAVE_FTW
1513#endif
1514
45e878bb 1515#ifndef HAVE_NFTW
173fc515 1516# define FTW_PHYS 1
45e878bb
NB
1517# ifndef HAVE_FTW
1518 struct FTW {};
173fc515
NB
1519# endif
1520#endif
1521
45e878bb
NB
1522#ifdef HAVE_FTW
1523# include <ftw.h>
1524#endif
1525
173fc515
NB
1526extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
1527
64c4757e 1528extern int Manage_ro(char *devname, int fd, int readonly);
d5a40416 1529extern int Manage_run(char *devname, int fd, struct context *c);
fe7e0e64
N
1530extern int Manage_stop(char *devname, int fd, int quiet,
1531 int will_retry);
64c4757e 1532extern int Manage_subdevs(char *devname, int fd,
833bb0f8 1533 struct mddev_dev *devlist, int verbose, int test,
f2e8393b 1534 enum update_opt update, int force);
1f48664b 1535extern int autodetect(void);
e5329c37 1536extern int Grow_Add_device(char *devname, int fd, char *newdev);
50f01ba5
N
1537extern int Grow_addbitmap(char *devname, int fd,
1538 struct context *c, struct shape *s);
32754b7d 1539extern int Grow_reshape(char *devname, int fd,
e2e53a2d 1540 struct mddev_dev *devlist,
32754b7d 1541 struct context *c, struct shape *s);
06b0d786 1542extern int Grow_restart(struct supertype *st, struct mdinfo *info,
ea0ebe96 1543 int *fdlist, int cnt, char *backup_file, int verbose);
e9e43ec3 1544extern int Grow_continue(int mdfd, struct supertype *st,
bdc2c569 1545 struct mdinfo *info, int forked, struct context *c);
860f11ed
AP
1546extern int Grow_consistency_policy(char *devname, int fd,
1547 struct context *c, struct shape *s);
b76b30e0 1548
3f54bd62
AK
1549extern int restore_backup(struct supertype *st,
1550 struct mdinfo *content,
1551 int working_disks,
1552 int spares,
54ded86f 1553 char **backup_filep,
3f54bd62 1554 int verbose);
bdc2c569 1555extern int Grow_continue_command(char *devname, int fd, struct context *c);
64c4757e 1556
7f91af49 1557extern int Assemble(struct supertype *st, char *mddev,
fa56eddb 1558 struct mddev_ident *ident,
87f26d14 1559 struct mddev_dev *devlist,
4977146a 1560 struct context *c);
64c4757e 1561
330c07f8
MT
1562extern int Build(struct mddev_ident *ident, struct mddev_dev *devlist, struct shape *s,
1563 struct context *c);
1564
1565extern int Create(struct supertype *st, struct mddev_ident *ident, int subdevs,
1566 struct mddev_dev *devlist, struct shape *s, struct context *c);
64c4757e 1567
ef898ce6 1568extern int Detail(char *dev, struct context *c);
9eafa1de 1569extern int Detail_Platform(struct superswitch *ss, int scan, int verbose, int export, char *controller_path);
e0d19036 1570extern int Query(char *dev);
6d388a88 1571extern int ExamineBadblocks(char *devname, int brief, struct supertype *forcest);
eec3f887
N
1572extern int Examine(struct mddev_dev *devlist, struct context *c,
1573 struct supertype *forcest);
a655e550 1574extern int Monitor(struct mddev_dev *devlist,
e0d19036 1575 char *mailaddr, char *alert_cmd,
95c50205
N
1576 struct context *c,
1577 int daemonise, int oneshot,
1578 int dosyslog, char *pidfile, int increments,
1579 int share);
64c4757e 1580
ba728be7
N
1581extern int Kill(char *dev, struct supertype *st, int force, int verbose, int noexcl);
1582extern int Kill_subarray(char *dev, char *subarray, int verbose);
f2e8393b 1583extern int Update_subarray(char *dev, char *subarray, enum update_opt update, struct mddev_ident *ident, int quiet);
b90c0e9a 1584extern int Wait(char *dev);
1ea04629 1585extern int WaitClean(char *dev, int verbose);
a740cf64 1586extern int SetAction(char *dev, char *action);
9a9dab36 1587
b11fe74d 1588extern int Incremental(struct mddev_dev *devlist, struct context *c,
11b6d91d 1589 struct supertype *st);
8382f19b 1590extern void RebuildMap(void);
d5a40416 1591extern int IncrementalScan(struct context *c, char *devnm);
950bc344 1592extern int IncrementalRemove(char *devname, char *path, int verbose);
c82f047c
NB
1593extern int CreateBitmap(char *filename, int force, char uuid[16],
1594 unsigned long chunksize, unsigned long daemon_sleep,
dfd4d8ee 1595 unsigned long write_behind,
dcec9ee5
NB
1596 unsigned long long array_size,
1597 int major);
55935d51 1598extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
9c030dad 1599extern int IsBitmapDirty(char *filename);
20b60dcd 1600extern int Write_rules(char *rule_name);
bf4fb153 1601extern int bitmap_update_uuid(int fd, int *uuid, int swap);
561ad559
AP
1602
1603/* calculate the size of the bitmap given the array size and bitmap chunksize */
1604static inline unsigned long long
1605bitmap_bits(unsigned long long array_size, unsigned long chunksize)
1606{
1607 return (array_size * 512 + chunksize - 1) / chunksize;
1608}
1609
74db60b0
N
1610extern int Dump_metadata(char *dev, char *dir, struct context *c,
1611 struct supertype *st);
1612extern int Restore_metadata(char *dev, char *dir, struct context *c,
1613 struct supertype *st, int only);
c82f047c 1614
9db2ab4e 1615int md_array_valid(int fd);
3ab8f4bf 1616int md_array_active(int fd);
a822017f 1617int md_array_is_active(struct mdinfo *info);
9cd39f01 1618int md_get_array_info(int fd, struct mdu_array_info_s *array);
018a4882 1619int md_set_array_info(int fd, struct mdu_array_info_s *array);
d97572f5 1620int md_get_disk_info(int fd, struct mdu_disk_info_s *disk);
e0d19036 1621extern int get_linux_version(void);
bc77ed53 1622extern int mdadm_version(char *version);
51d4261c 1623extern unsigned long long parse_size(char *size);
64c4757e 1624extern int parse_uuid(char *str, int uuid[4]);
913f07d1 1625int default_layout(struct supertype *st, int level, int verbose);
5339f996 1626extern int is_near_layout_10(int layout);
4a06e2c2
N
1627extern int parse_layout_10(char *layout);
1628extern int parse_layout_faulty(char *layout);
25aa7329 1629extern int parse_num(int *dest, const char *num);
4de90913 1630extern int parse_cluster_confirm_arg(char *inp, char **devname, int *slot);
682c7051
NB
1631extern int check_ext2(int fd, char *name);
1632extern int check_reiser(int fd, char *name);
1633extern int check_raid(int fd, char *name);
53ed6ac3
KW
1634extern int check_partitions(int fd, char *dname,
1635 unsigned long long freesize,
1636 unsigned long long size);
e2eb503b 1637extern bool is_name_posix_compatible(const char *path);
0a6bff09 1638extern int fstat_is_blkdev(int fd, char *devname, dev_t *rdev);
9e04ac1c 1639extern int stat_is_blkdev(char *devname, dev_t *rdev);
64c4757e 1640
67417d92 1641extern bool is_string_lq(const char * const str, size_t max_len);
a35aa68f 1642extern bool is_dev_alive(char *path);
dd0781e5 1643extern int get_mdp_major(void);
cc742d38 1644extern int get_maj_min(char *dev, int *major, int *minor);
7211116c 1645extern bool is_bit_set(int *val, unsigned char index);
8b0dabea 1646extern int dev_open(char *dev, int flags);
4dd2df09 1647extern int open_dev(char *devnm);
8832342d 1648extern void reopen_mddev(int mdfd);
4dd2df09
N
1649extern int open_dev_flags(char *devnm, int flags);
1650extern int open_dev_excl(char *devnm);
8d80900b 1651extern int is_standard(char *dev, int *nump);
9008ed1c 1652extern int same_dev(char *one, char *two);
9eafa1de 1653extern int compare_paths (char* path1,char* path2);
a7dec3fd 1654extern void enable_fds(int devices);
ff6bb131 1655extern void manage_fork_fds(int close_all);
723d1df4 1656extern int continue_via_systemd(char *devnm, char *service_name, char *prefix);
dd0781e5 1657
7fcbfd7c 1658extern void ident_init(struct mddev_ident *ident);
ae5f13a9 1659extern mdadm_status_t ident_set_devname(struct mddev_ident *ident, const char *devname);
67417d92 1660extern mdadm_status_t ident_set_name(struct mddev_ident *ident, const char *name);
7fcbfd7c 1661
f1ae21c4 1662extern int parse_auto(char *str, char *msg, int config);
fa56eddb 1663extern struct mddev_ident *conf_get_ident(char *dev);
a655e550 1664extern struct mddev_dev *conf_get_devs(void);
8382f19b 1665extern int conf_test_dev(char *devname);
4e8d9f0a 1666extern int conf_test_metadata(const char *version, struct dev_policy *pol, int is_homehost);
8aec876d
NB
1667extern struct createinfo *conf_get_create_info(void);
1668extern void set_conffile(char *file);
1669extern char *conf_get_mailaddr(void);
1670extern char *conf_get_mailfrom(void);
1671extern char *conf_get_program(void);
0ac91628 1672extern char *conf_get_homehost(int *require_homehostp);
7716570e 1673extern char *conf_get_homecluster(void);
0663137c 1674extern int conf_get_monitor_delay(void);
336e13fc 1675extern bool conf_get_sata_opal_encryption_no_verify(void);
e0d19036 1676extern char *conf_line(FILE *file);
98c6faba 1677extern char *conf_word(FILE *file, int allow_key);
7103b9b8 1678extern void print_quoted(char *str);
06d2ffc3 1679extern int use_udev(void);
9935cf0f 1680extern void print_escape(char *str);
2eba8496 1681extern unsigned long GCD(unsigned long a, unsigned long b);
0ac91628 1682extern int conf_name_is_free(char *name);
ae5f13a9 1683extern bool is_devname_ignore(const char *devname);
25aa7329
MT
1684extern bool is_devname_md_numbered(const char *devname);
1685extern bool is_devname_md_d_numbered(const char *devname);
7c336758 1686extern int conf_verify_devnames(struct mddev_ident *array_list);
0ac91628 1687extern int devname_matches(char *name, char *match);
2244d1a9
N
1688extern struct mddev_ident *conf_match(struct supertype *st,
1689 struct mdinfo *info,
1690 char *devname,
1691 int verbose, int *rvp);
0ac91628 1692
e0d19036
NB
1693extern void free_line(char *line);
1694extern int match_oneof(char *devices, char *devname);
e0d19036 1695extern void uuid_from_super(int uuid[4], mdp_super_t *super);
22e263f6 1696extern const int uuid_zero[4];
f277ce36 1697extern int same_uuid(int a[4], int b[4], int swapuuid);
350f29f9 1698extern void copy_uuid(void *a, int b[4], int swapuuid);
aae5a112 1699extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
ba65d917 1700extern char *fname_from_uuid(struct mdinfo *info, char *buf);
4b1ac34b 1701extern unsigned long calc_csum(void *super, int bytes);
583315d9 1702extern int enough(int level, int raid_disks, int layout, int clean,
de5a472e 1703 char *avail);
e0d19036 1704extern int ask(char *mesg);
8fac0577 1705extern unsigned long long get_component_size(int fd);
0430ed48 1706extern void remove_partitions(int fd);
691c6ee1 1707extern int test_partition(int fd);
bfd76b93 1708extern int test_partition_from_id(dev_t id);
577e8448 1709extern int get_data_disks(int level, int layout, int raid_disks);
5f8097be
NB
1710extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
1711 int chunksize, unsigned long long devsize);
edd8d13c
NB
1712extern int flush_metadata_updates(struct supertype *st);
1713extern void append_metadata_update(struct supertype *st, void *buf, int len);
6234c63c 1714extern int assemble_container_content(struct supertype *st, int mdfd,
11b6d91d
N
1715 struct mdinfo *content,
1716 struct context *c,
9ca39acb
N
1717 char *chosen_name, int *result);
1718#define INCR_NO 1
1719#define INCR_UNSAFE 2
1720#define INCR_ALREADY 4
1721#define INCR_YES 8
0c0f09cb 1722
51a9f2fc
MT
1723extern bool devid_matches_criteria(struct supertype *st, dev_t devid, struct spare_criteria *sc);
1724extern bool disk_fd_matches_criteria(struct supertype *st, int disk_fd, struct spare_criteria *sc);
326727d9 1725extern struct mdinfo *container_choose_spares(struct supertype *st,
fbfdcb06 1726 struct spare_criteria *criteria,
326727d9
AC
1727 struct domainlist *domlist,
1728 char *spare_group,
1729 const char *metadata, int get_one);
d52bb542 1730extern int move_spare(char *from_devname, char *to_devname, dev_t devid);
7801ac20
N
1731extern int add_disk(int mdfd, struct supertype *st,
1732 struct mdinfo *sra, struct mdinfo *info);
de6ae750
N
1733extern int remove_disk(int mdfd, struct supertype *st,
1734 struct mdinfo *sra, struct mdinfo *info);
1ab9ed2a
N
1735extern int hot_remove_disk(int mdfd, unsigned long dev, int force);
1736extern int sys_hot_remove_disk(int statefd, int force);
f35f2525 1737extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
1e5c6983 1738unsigned long long min_recovery_start(struct mdinfo *array);
cd29a5c8 1739
56eb10c0 1740extern char *human_size(long long bytes);
f0ec6710 1741extern char *human_size_brief(long long bytes, int prefix);
e4965ef8 1742extern void print_r10_layout(int layout);
e0d19036 1743
4dd2df09 1744extern char *find_free_devnm(int use_partitions);
63152c1b 1745
e0d19036 1746extern void put_md_name(char *name);
d3c40fab 1747extern char *devid2kname(dev_t devid);
13db17bd
ML
1748extern char *devid2devnm(dev_t devid);
1749extern dev_t devnm2devid(char *devnm);
4dd2df09 1750extern char *get_md_name(char *devnm);
e0d19036 1751
c913b90e 1752extern char DefaultConfFile[];
b5e64645 1753
69207ff6 1754extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
cd6cbb08 1755 char *chosen, int block_udev);
69207ff6
N
1756/* values for 'trustworthy' */
1757#define LOCAL 1
d1d3482b 1758#define LOCAL_ANY 10
69207ff6
N
1759#define FOREIGN 2
1760#define METADATA 3
6be1d39d 1761extern int open_mddev(char *dev, int report_errors);
e702f392 1762extern int is_mddev(char *dev);
a322f70c 1763extern int open_container(int fd);
4dd2df09
N
1764extern int metadata_container_matches(char *metadata, char *devnm);
1765extern int metadata_subdev_matches(char *metadata, char *devnm);
33414a01
DW
1766extern int is_container_member(struct mdstat_ent *ent, char *devname);
1767extern int is_subarray_active(char *subarray, char *devname);
feab51f8 1768extern int open_subarray(char *dev, char *subarray, struct supertype *st, int quiet);
33414a01 1769extern struct superswitch *version_to_superswitch(char *vers);
b5e64645 1770
4dd2df09
N
1771extern int mdmon_running(char *devnm);
1772extern int mdmon_pid(char *devnm);
40ebbb9c 1773extern int check_env(char *name);
148acb7b 1774extern __u32 random32(void);
c5f71c24 1775extern void random_uuid(__u8 *buf);
4dd2df09 1776extern int start_mdmon(char *devnm);
a931db9e 1777
999b4972
N
1778extern int child_monitor(int afd, struct mdinfo *sra, struct reshape *reshape,
1779 struct supertype *st, unsigned long stripes,
1780 int *fds, unsigned long long *offsets,
1781 int dests, int *destfd, unsigned long long *destoffsets);
10f22854 1782void abort_reshape(struct mdinfo *sra);
999b4972 1783
afa368f4
N
1784void *super1_make_v0(struct supertype *st, struct mdinfo *info, mdp_super_t *sb0);
1785
f96b1302
AP
1786extern char *stat2kname(struct stat *st);
1787extern char *fd2kname(int fd);
4dd2df09 1788extern char *stat2devnm(struct stat *st);
27ad4900 1789bool stat_is_md_dev(struct stat *st);
4dd2df09 1790extern char *fd2devnm(int fd);
a931db9e 1791
a9c15847 1792extern int in_initrd(void);
d15a1f72 1793
e80357f8
GJ
1794struct cmap_hooks {
1795 void *cmap_handle; /* corosync lib related */
1796
1797 int (*initialize)(cmap_handle_t *handle);
1798 int (*get_string)(cmap_handle_t handle,
1799 const char *string,
1800 char **name);
1801 int (*finalize)(cmap_handle_t handle);
1802};
1803
1804extern void set_cmap_hooks(void);
1805extern void set_hooks(void);
1806
d15a1f72
GJ
1807struct dlm_hooks {
1808 void *dlm_handle; /* dlm lib related */
1809
1810 dlm_lshandle_t (*create_lockspace)(const char *name,
1811 unsigned int mode);
1b7eb962 1812 dlm_lshandle_t (*open_lockspace)(const char *name);
d15a1f72
GJ
1813 int (*release_lockspace)(const char *name, dlm_lshandle_t ls,
1814 int force);
1815 int (*ls_lock)(dlm_lshandle_t lockspace, uint32_t mode,
1816 struct dlm_lksb *lksb, uint32_t flags,
1817 const void *name, unsigned int namelen,
1818 uint32_t parent, void (*astaddr) (void *astarg),
1819 void *astarg, void (*bastaddr) (void *astarg),
1820 void *range);
1b7eb962
GJ
1821 int (*ls_unlock_wait)(dlm_lshandle_t lockspace, uint32_t lkid,
1822 uint32_t flags, struct dlm_lksb *lksb);
d15a1f72
GJ
1823 int (*ls_get_fd)(dlm_lshandle_t ls);
1824 int (*dispatch)(int fd);
1825};
1826
7716570e 1827extern int get_cluster_name(char **name);
81a8a694 1828extern int dlm_funs_ready(void);
1b7eb962
GJ
1829extern int cluster_get_dlmlock(void);
1830extern int cluster_release_dlmlock(void);
d15a1f72 1831extern void set_dlm_hooks(void);
a9c15847 1832
239b3cc0
MG
1833#define MSEC_TO_NSEC(msec) ((msec) * 1000000)
1834#define USEC_TO_NSEC(usec) ((usec) * 1000)
1835extern void sleep_for(unsigned int sec, long nsec, bool wake_after_interrupt);
ee9dcf95
MG
1836extern bool is_directory(const char *path);
1837extern bool is_file(const char *path);
21e622f2 1838extern int s_gethostname(char *buf, int buf_len);
239b3cc0 1839
2d762ade 1840#define _ROUND_UP(val, base) (((val) + (base) - 1) & ~(base - 1))
de897065
JS
1841#define ROUND_UP(val, base) _ROUND_UP(val, (typeof(val))(base))
1842#define ROUND_UP_PTR(ptr, base) ((typeof(ptr)) \
1843 (ROUND_UP((unsigned long)(ptr), base)))
6416d527 1844
3c558363
N
1845static inline int is_subarray(char *vers)
1846{
e9dd1598 1847 /* The version string for a 'subarray' (an array in a container)
1011e834 1848 * is
e9dd1598 1849 * /containername/componentname for normal read-write arrays
746a6567
N
1850 * -containername/componentname for arrays which mdmon must not
1851 * reconfigure. They might be read-only
1852 * or might be undergoing reshape etc.
e9dd1598
N
1853 * containername is e.g. md0, md_d1
1854 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
1855 */
1856 return (*vers == '/' || *vers == '-');
3c558363
N
1857}
1858
e5408a32
DW
1859static inline char *to_subarray(struct mdstat_ent *ent, char *container)
1860{
1861 return &ent->metadata_version[10+strlen(container)+1];
1862}
1863
83a379cf
LF
1864/**
1865 * signal_s() - Wrapper for sigaction() with signal()-like interface.
1866 * @sig: The signal to set the signal handler to.
1867 * @handler: The signal handler.
1868 *
1869 * Return: previous handler or SIG_ERR on failure.
1870 */
1871static inline sighandler_t signal_s(int sig, sighandler_t handler)
1872{
38cb95dd
MK
1873 struct sigaction new_act = {0};
1874 struct sigaction old_act = {0};
83a379cf
LF
1875
1876 new_act.sa_handler = handler;
83a379cf
LF
1877
1878 if (sigaction(sig, &new_act, &old_act) == 0)
1879 return old_act.sa_handler;
1880
1881 return SIG_ERR;
1882}
1883
2a24d7b6
DW
1884#ifdef DEBUG
1885#define dprintf(fmt, arg...) \
1ade5cc1
N
1886 fprintf(stderr, "%s: %s: "fmt, Name, __func__, ##arg)
1887#define dprintf_cont(fmt, arg...) \
2a24d7b6
DW
1888 fprintf(stderr, fmt, ##arg)
1889#else
1890#define dprintf(fmt, arg...) \
1ade5cc1
N
1891 ({ if (0) fprintf(stderr, "%s: %s: " fmt, Name, __func__, ##arg); 0; })
1892#define dprintf_cont(fmt, arg...) \
2a24d7b6
DW
1893 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
1894#endif
7fcbfd7c 1895
1a0ee0ba
DK
1896static inline int xasprintf(char **strp, const char *fmt, ...) {
1897 va_list ap;
1898 int ret;
1899 va_start(ap, fmt);
089485cb 1900 ret = vasprintf(strp, fmt, ap);
1a0ee0ba
DK
1901 va_end(ap);
1902 assert(ret >= 0);
1903 return ret;
1904}
2a24d7b6 1905
1ade5cc1
N
1906#ifdef DEBUG
1907#define pr_err(fmt, args...) fprintf(stderr, "%s: %s: "fmt, Name, __func__, ##args)
1908#else
d56dd607 1909#define pr_err(fmt, args...) fprintf(stderr, "%s: "fmt, Name, ##args)
1ade5cc1 1910#endif
e7b84f9d
N
1911#define cont_err(fmt ...) fprintf(stderr, " " fmt)
1912
9364dbfb
LG
1913#define pr_info(fmt, args...) printf("%s: "fmt, Name, ##args)
1914
cc75b0fa
BK
1915#define pr_vrb(fmt, arg...) ((void)(verbose && pr_err(fmt, ##arg)))
1916
503975b9
N
1917void *xmalloc(size_t len);
1918void *xrealloc(void *ptr, size_t len);
1919void *xcalloc(size_t num, size_t size);
1920char *xstrdup(const char *str);
1921
b5e64645
NB
1922#define LEVEL_MULTIPATH (-4)
1923#define LEVEL_LINEAR (-1)
1924#define LEVEL_FAULTY (-5)
1925
17f25ca6
NB
1926/* kernel module doesn't know about these */
1927#define LEVEL_CONTAINER (-100)
a322f70c 1928#define LEVEL_UNSUPPORTED (-200)
17f25ca6 1929
56cb05c4
N
1930/* the kernel does know about this one ... */
1931#define LEVEL_NONE (-1000000)
1932
b5e64645
NB
1933/* faulty stuff */
1934
1935#define WriteTransient 0
1936#define ReadTransient 1
1937#define WritePersistent 2
1938#define ReadPersistent 3
1939#define WriteAll 4 /* doesn't go to device */
1940#define ReadFixable 5
1941#define Modes 6
1942
1943#define ClearErrors 31
1944#define ClearFaults 30
1945
1946#define AllPersist 100 /* internal use only */
1947#define NoPersist 101
1948
1949#define ModeMask 0x1f
1950#define ModeShift 5
fe6729fa 1951
fe6729fa
NB
1952#ifdef __TINYC__
1953#undef minor
1954#undef major
1955#undef makedev
1956#define minor(x) ((x)&0xff)
1957#define major(x) (((x)>>8)&0xff)
1958#define makedev(M,m) (((M)<<8) | (m))
1959#endif
1960
329dfc28
N
1961enum r0layout {
1962 RAID0_ORIG_LAYOUT = 1,
1963 RAID0_ALT_MULTIZONE_LAYOUT = 2,
1964};
1965
b640a252 1966/* for raid4/5/6 */
e86c9dd6
NB
1967#define ALGORITHM_LEFT_ASYMMETRIC 0
1968#define ALGORITHM_RIGHT_ASYMMETRIC 1
1969#define ALGORITHM_LEFT_SYMMETRIC 2
1970#define ALGORITHM_RIGHT_SYMMETRIC 3
b640a252
N
1971
1972/* Define non-rotating (raid4) algorithms. These allow
1973 * conversion of raid4 to raid5.
1974 */
1975#define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
1976#define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
1977
1978/* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
1979 * Firstly, the exact positioning of the parity block is slightly
1980 * different between the 'LEFT_*' modes of md and the "_N_*" modes
1981 * of DDF.
1982 * Secondly, or order of datablocks over which the Q syndrome is computed
1983 * is different.
1984 * Consequently we have different layouts for DDF/raid6 than md/raid6.
1985 * These layouts are from the DDFv1.2 spec.
1986 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
1987 * leaves RLQ=3 as 'Vendor Specific'
1988 */
1989
1990#define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
1991#define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
1992#define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
1993
b640a252
N
1994/* For every RAID5 algorithm we define a RAID6 algorithm
1995 * with exactly the same layout for data and parity, and
1996 * with the Q block always on the last device (N-1).
1997 * This allows trivial conversion from RAID5 to RAID6
1998 */
1999#define ALGORITHM_LEFT_ASYMMETRIC_6 16
2000#define ALGORITHM_RIGHT_ASYMMETRIC_6 17
2001#define ALGORITHM_LEFT_SYMMETRIC_6 18
2002#define ALGORITHM_RIGHT_SYMMETRIC_6 19
2003#define ALGORITHM_PARITY_0_6 20
2004#define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
2005
33a6535d
AW
2006/* Define PATH_MAX in case we don't use glibc or standard library does
2007 * not have PATH_MAX defined. Assume max path length is 4K characters.
2008 */
2009#ifndef PATH_MAX
2010#define PATH_MAX 4096
2011#endif
2012
9dad51d4
N
2013#define RESYNC_NONE -1
2014#define RESYNC_DELAYED -2
2015#define RESYNC_PENDING -3
1c294b5d
LZ
2016#define RESYNC_REMOTE -4
2017#define RESYNC_UNKNOWN -5
a0963a86 2018
480f3566
N
2019/* When using "GET_DISK_INFO" it isn't certain how high
2020 * we need to check. So we impose an absolute limit of
2021 * MAX_DISKS. This needs to be much more than the largest
2022 * number of devices any metadata can support. Currently
2023 * v1.x can support 1920
2024 */
2025#define MAX_DISKS 4096
2026
d04f65f4
N
2027/* Sometimes the 'size' value passed needs to mean "Maximum".
2028 * In those cases with use MAX_SIZE
2029 */
2030#define MAX_SIZE 1
822e393a
N
2031
2032/* We want to use unsigned numbers for sector counts, but need
2033 * a value for 'invalid'. Use '1'.
2034 */
2035#define INVALID_SECTORS 1
72ca9bcf
N
2036/* And another special number needed for --data_offset=variable */
2037#define VARIABLE_OFFSET 3
ced5fa8b 2038
6f2af6a4
KT
2039/**
2040 * is_container() - check if @level is &LEVEL_CONTAINER
2041 * @level: level value
2042 *
2043 * return:
2044 * 1 if level is equal to &LEVEL_CONTAINER, 0 otherwise.
2045 */
2046static inline int is_container(const int level)
2047{
2048 if (level == LEVEL_CONTAINER)
2049 return 1;
2050 return 0;
2051}
b823c8f9
MK
2052
2053#define STR_COMMON_NONE "none"
2054
2055/**
2056 * str_is_none() - check if @str starts with "none".
2057 * @str: string
2058 *
2059 * return:
2060 * true if string starts with "none", false otherwise.
2061 */
2062static inline bool str_is_none(char *str)
2063{
2064 if (strncmp(str, STR_COMMON_NONE, sizeof(STR_COMMON_NONE) - 1) == 0)
2065 return true;
2066 return false;
2067}