]> git.ipfire.org Git - thirdparty/mdadm.git/blob - mdadm.h
9cf15c41b1b0c1a483a01d9f48bbc54b1e473a1b
[thirdparty/mdadm.git] / mdadm.h
1 /*
2 * mdadm - manage Linux "md" devices aka RAID arrays.
3 *
4 * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
5 *
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 * Author: Neil Brown
22 * Email: <neilb@suse.de>
23 */
24
25 #define _GNU_SOURCE
26 #define _FILE_OFFSET_BITS 64
27 #include <unistd.h>
28 #if !defined(__dietlibc__) && !defined(__KLIBC__)
29 extern __off64_t lseek64 __P ((int __fd, __off64_t __offset, int __whence));
30 #else
31 # if defined(__NO_STAT64) || __WORDSIZE != 32
32 # define lseek64 lseek
33 # endif
34 #endif
35
36 #include <sys/types.h>
37 #include <sys/stat.h>
38 #include <stdlib.h>
39 #include <time.h>
40 #include <sys/time.h>
41 #include <getopt.h>
42 #include <fcntl.h>
43 #include <stdio.h>
44 #include <errno.h>
45 #include <string.h>
46 #include <syslog.h>
47 #ifdef __dietlibc__
48 #include <strings.h>
49 /* dietlibc has deprecated random and srandom!! */
50 #define random rand
51 #define srandom srand
52 #endif
53
54
55 #include <linux/kdev_t.h>
56 /*#include <linux/fs.h> */
57 #include <sys/mount.h>
58 #include <asm/types.h>
59 #include <sys/ioctl.h>
60 #define MD_MAJOR 9
61 #define MdpMinorShift 6
62
63 #ifndef BLKGETSIZE64
64 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
65 #endif
66
67 #define DEFAULT_BITMAP_CHUNK 4096
68 #define DEFAULT_BITMAP_DELAY 5
69 #define DEFAULT_MAX_WRITE_BEHIND 256
70
71 #include "md_u.h"
72 #include "md_p.h"
73 #include "bitmap.h"
74 #include "msg.h"
75
76 #include <endian.h>
77 /* Redhat don't like to #include <asm/byteorder.h>, and
78 * some time include <linux/byteorder/xxx_endian.h> isn't enough,
79 * and there is no standard conversion function so... */
80 /* And dietlibc doesn't think byteswap is ok, so.. */
81 /* #include <byteswap.h> */
82 #define bswap_16(x) (((x) & 0x00ffU) << 8 | \
83 ((x) & 0xff00U) >> 8)
84 #define bswap_32(x) (((x) & 0x000000ffU) << 24 | \
85 ((x) & 0xff000000U) >> 24 | \
86 ((x) & 0x0000ff00U) << 8 | \
87 ((x) & 0x00ff0000U) >> 8)
88 #define bswap_64(x) (((x) & 0x00000000000000ffULL) << 56 | \
89 ((x) & 0xff00000000000000ULL) >> 56 | \
90 ((x) & 0x000000000000ff00ULL) << 40 | \
91 ((x) & 0x00ff000000000000ULL) >> 40 | \
92 ((x) & 0x0000000000ff0000ULL) << 24 | \
93 ((x) & 0x0000ff0000000000ULL) >> 24 | \
94 ((x) & 0x00000000ff000000ULL) << 8 | \
95 ((x) & 0x000000ff00000000ULL) >> 8)
96
97 #if !defined(__KLIBC__)
98 #if BYTE_ORDER == LITTLE_ENDIAN
99 #define __cpu_to_le16(_x) (_x)
100 #define __cpu_to_le32(_x) (_x)
101 #define __cpu_to_le64(_x) (_x)
102 #define __le16_to_cpu(_x) (_x)
103 #define __le32_to_cpu(_x) (_x)
104 #define __le64_to_cpu(_x) (_x)
105
106 #define __cpu_to_be16(_x) bswap_16(_x)
107 #define __cpu_to_be32(_x) bswap_32(_x)
108 #define __cpu_to_be64(_x) bswap_64(_x)
109 #define __be16_to_cpu(_x) bswap_16(_x)
110 #define __be32_to_cpu(_x) bswap_32(_x)
111 #define __be64_to_cpu(_x) bswap_64(_x)
112 #elif BYTE_ORDER == BIG_ENDIAN
113 #define __cpu_to_le16(_x) bswap_16(_x)
114 #define __cpu_to_le32(_x) bswap_32(_x)
115 #define __cpu_to_le64(_x) bswap_64(_x)
116 #define __le16_to_cpu(_x) bswap_16(_x)
117 #define __le32_to_cpu(_x) bswap_32(_x)
118 #define __le64_to_cpu(_x) bswap_64(_x)
119
120 #define __cpu_to_be16(_x) (_x)
121 #define __cpu_to_be32(_x) (_x)
122 #define __cpu_to_be64(_x) (_x)
123 #define __be16_to_cpu(_x) (_x)
124 #define __be32_to_cpu(_x) (_x)
125 #define __be64_to_cpu(_x) (_x)
126 #else
127 # error "unknown endianness."
128 #endif
129 #endif /* __KLIBC__ */
130
131
132
133 /* general information that might be extracted from a superblock */
134 struct mdinfo {
135 mdu_array_info_t array;
136 mdu_disk_info_t disk;
137 __u64 events;
138 int uuid[4];
139 char name[33];
140 unsigned long long data_offset;
141 unsigned long long component_size; /* same as array.size, except in
142 * sectors and up to 64bits.
143 */
144 unsigned long long custom_array_size; /* size for non-default sized
145 * arrays (in sectors)
146 */
147 int reshape_active;
148 unsigned long long reshape_progress;
149 union {
150 unsigned long long resync_start; /* per-array resync position */
151 unsigned long long recovery_start; /* per-device rebuild position */
152 #define MaxSector (~0ULL) /* resync/recovery complete position */
153 };
154 unsigned long safe_mode_delay; /* ms delay to mark clean */
155 int new_level, delta_disks, new_layout, new_chunk;
156 int errors;
157 int cache_size; /* size of raid456 stripe cache*/
158 int mismatch_cnt;
159 char text_version[50];
160 void *update_private; /* for passing metadata-format
161 * specific update data
162 * between successive calls to
163 * update_super()
164 */
165
166 int container_member; /* for assembling external-metatdata arrays
167 * This is to be used internally by metadata
168 * handler only */
169
170 char sys_name[20];
171 struct mdinfo *devs;
172 struct mdinfo *next;
173
174 /* Device info for mdmon: */
175 int recovery_fd;
176 int state_fd;
177 #define DS_FAULTY 1
178 #define DS_INSYNC 2
179 #define DS_WRITE_MOSTLY 4
180 #define DS_SPARE 8
181 #define DS_BLOCKED 16
182 #define DS_REMOVE 1024
183 #define DS_UNBLOCK 2048
184 int prev_state, curr_state, next_state;
185
186 };
187
188 struct createinfo {
189 int uid;
190 int gid;
191 int autof;
192 int mode;
193 int symlinks;
194 struct supertype *supertype;
195 };
196
197 #define Name "mdadm"
198
199 enum mode {
200 ASSEMBLE=1,
201 BUILD,
202 CREATE,
203 MANAGE,
204 MISC,
205 MONITOR,
206 GROW,
207 INCREMENTAL,
208 AUTODETECT,
209 };
210
211 extern char short_options[];
212 extern char short_bitmap_options[];
213 extern char short_bitmap_auto_options[];
214 extern struct option long_options[];
215 extern char Version[], Usage[], Help[], OptionHelp[],
216 Help_create[], Help_build[], Help_assemble[], Help_grow[],
217 Help_incr[],
218 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
219
220 /* for option that don't have short equivilents, we assign arbitrary
221 * small numbers. '1' means an undecorated option, so we start at '2'.
222 */
223 enum special_options {
224 AssumeClean = 2,
225 BitmapChunk,
226 WriteBehind,
227 ReAdd,
228 NoDegraded,
229 Sparc22,
230 BackupFile,
231 HomeHost,
232 AutoHomeHost,
233 Symlinks,
234 AutoDetect,
235 Waitclean,
236 DetailPlatform,
237 };
238
239 /* structures read from config file */
240 /* List of mddevice names and identifiers
241 * Identifiers can be:
242 * uuid=128-hex-uuid
243 * super-minor=decimal-minor-number-from-superblock
244 * devices=comma,separated,list,of,device,names,with,wildcards
245 *
246 * If multiple fields are present, the intersection of all matching
247 * devices is considered
248 */
249 #define UnSet (0xfffe)
250 typedef struct mddev_ident_s {
251 char *devname;
252
253 int uuid_set;
254 int uuid[4];
255 char name[33];
256
257 unsigned int super_minor;
258
259 char *devices; /* comma separated list of device
260 * names with wild cards
261 */
262 int level;
263 unsigned int raid_disks;
264 unsigned int spare_disks;
265 struct supertype *st;
266 int autof; /* 1 for normal, 2 for partitioned */
267 char *spare_group;
268 char *bitmap_file;
269 int bitmap_fd;
270
271 char *container; /* /dev/whatever name of container, or
272 * uuid of container. You would expect
273 * this to be the 'devname' or UUID
274 * of some other entry.
275 */
276 char *member; /* subarray within a container */
277
278 struct mddev_ident_s *next;
279 } *mddev_ident_t;
280
281 /* List of device names - wildcards expanded */
282 typedef struct mddev_dev_s {
283 char *devname;
284 char disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
285 * Not set for names read from .config
286 */
287 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
288 char re_add;
289 char used; /* set when used */
290 struct mdinfo *content; /* If devname is a container, this might list
291 * the remaining member arrays. */
292 struct mddev_dev_s *next;
293 } *mddev_dev_t;
294
295 typedef struct mapping {
296 char *name;
297 int num;
298 } mapping_t;
299
300
301 struct mdstat_ent {
302 char *dev;
303 int devnum;
304 int active;
305 char *level;
306 char *pattern; /* U or up, _ for down */
307 int percent; /* -1 if no resync */
308 int resync; /* 1 if resync, 0 if recovery */
309 int devcnt;
310 int raid_disks;
311 int chunk_size;
312 char * metadata_version;
313 struct mdstat_ent *next;
314 };
315
316 extern struct mdstat_ent *mdstat_read(int hold, int start);
317 extern void free_mdstat(struct mdstat_ent *ms);
318 extern void mdstat_wait(int seconds);
319 extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
320 extern int mddev_busy(int devnum);
321
322 struct map_ent {
323 struct map_ent *next;
324 int devnum;
325 char metadata[20];
326 int uuid[4];
327 int bad;
328 char *path;
329 };
330 extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
331 int uuid[4], char *path);
332 extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
333 extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
334 extern struct map_ent *map_by_name(struct map_ent **map, char *name);
335 extern void map_read(struct map_ent **melp);
336 extern int map_write(struct map_ent *mel);
337 extern void map_delete(struct map_ent **mapp, int devnum);
338 extern void map_free(struct map_ent *map);
339 extern void map_add(struct map_ent **melp,
340 int devnum, char *metadata, int uuid[4], char *path);
341 extern int map_lock(struct map_ent **melp);
342 extern void map_unlock(struct map_ent **melp);
343
344 /* various details can be requested */
345 enum sysfs_read_flags {
346 GET_LEVEL = (1 << 0),
347 GET_LAYOUT = (1 << 1),
348 GET_COMPONENT = (1 << 2),
349 GET_CHUNK = (1 << 3),
350 GET_CACHE = (1 << 4),
351 GET_MISMATCH = (1 << 5),
352 GET_VERSION = (1 << 6),
353 GET_DISKS = (1 << 7),
354 GET_DEGRADED = (1 << 8),
355 GET_SAFEMODE = (1 << 9),
356 GET_DEVS = (1 << 10), /* gets role, major, minor */
357 GET_OFFSET = (1 << 11),
358 GET_SIZE = (1 << 12),
359 GET_STATE = (1 << 13),
360 GET_ERROR = (1 << 14),
361 SKIP_GONE_DEVS = (1 << 15),
362 };
363
364 /* If fd >= 0, get the array it is open on,
365 * else use devnum. >=0 -> major9. <0.....
366 */
367 extern int sysfs_open(int devnum, char *devname, char *attr);
368 extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
369 extern void sysfs_free(struct mdinfo *sra);
370 extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
371 extern int sysfs_attr_match(const char *attr, const char *str);
372 extern int sysfs_match_word(const char *word, char **list);
373 extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
374 char *name, char *val);
375 extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
376 char *name, unsigned long long val);
377 extern int sysfs_uevent(struct mdinfo *sra, char *event);
378 extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
379 char *name);
380 extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
381 extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
382 char *name, unsigned long long *val);
383 extern int sysfs_fd_get_str(int fd, char *val, int size);
384 extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
385 char *name, char *val, int size);
386 extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
387 extern int sysfs_set_array(struct mdinfo *info, int vers);
388 extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd,
389 int in_sync);
390 extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
391 extern int sysfs_unique_holder(int devnum, long rdev);
392 extern int load_sys(char *path, char *buf);
393
394
395 extern int save_stripes(int *source, unsigned long long *offsets,
396 int raid_disks, int chunk_size, int level, int layout,
397 int nwrites, int *dest,
398 unsigned long long start, unsigned long long length,
399 char *buf);
400 extern int restore_stripes(int *dest, unsigned long long *offsets,
401 int raid_disks, int chunk_size, int level, int layout,
402 int source, unsigned long long read_offset,
403 unsigned long long start, unsigned long long length);
404
405 #ifndef Sendmail
406 #define Sendmail "/usr/lib/sendmail -t"
407 #endif
408
409 #define SYSLOG_FACILITY LOG_DAEMON
410
411 extern char *map_num(mapping_t *map, int num);
412 extern int map_name(mapping_t *map, char *name);
413 extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
414
415 extern char *map_dev(int major, int minor, int create);
416
417 struct active_array;
418 struct metadata_update;
419
420 /* A superswitch provides entry point the a metadata handler.
421 *
422 * The super_switch primarily operates on some "metadata" that
423 * is accessed via the 'supertype'.
424 * This metadata has one of three possible sources.
425 * 1/ It is read from a single device. In this case it may not completely
426 * describe the array or arrays as some information might be on other
427 * devices.
428 * 2/ It is read from all devices in a container. In this case all
429 * information is present.
430 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
431 * be complete once enough ->add_to_super calls have completed.
432 *
433 * When creating an array inside a container, the metadata will be
434 * formed by a combination of 2 and 3. The metadata or the array is read,
435 * then new information is added.
436 *
437 * The metadata must sometimes have a concept of a 'current' array
438 * and a 'current' device.
439 * The 'current' array is set by init_super to be the newly created array,
440 * or is set by super_by_fd when it finds it is looking at an array inside
441 * a container.
442 *
443 * The 'current' device is either the device that the metadata was read from
444 * in case 1, or the last device added by add_to_super in case 3.
445 * Case 2 does not identify a 'current' device.
446 */
447 extern struct superswitch {
448
449 /* Used to report details of metadata read from a component
450 * device. ->load_super has been called.
451 */
452 void (*examine_super)(struct supertype *st, char *homehost);
453 void (*brief_examine_super)(struct supertype *st, int verbose);
454 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
455 void (*export_examine_super)(struct supertype *st);
456
457 /* Used to report details of an active array.
458 * ->load_super was possibly given a 'component' string.
459 */
460 void (*detail_super)(struct supertype *st, char *homehost);
461 void (*brief_detail_super)(struct supertype *st);
462 void (*export_detail_super)(struct supertype *st);
463
464 /* Optional: platform hardware / firmware details */
465 int (*detail_platform)(int verbose, int enumerate_only);
466
467 /* Used:
468 * to get uuid to storing in bitmap metadata
469 * and 'reshape' backup-data metadata
470 * To see if a device is being re-added to an array it was part of.
471 */
472 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
473
474 /* Extract generic details from metadata. This could be details about
475 * the container, or about an individual array within the container.
476 * The determination is made either by:
477 * load_super being given a 'component' string.
478 * validate_geometry determining what to create.
479 * The info includes both array information and device information.
480 * The particular device should be:
481 * The last device added by add_to_super
482 * The device the metadata was loaded from by load_super
483 */
484 void (*getinfo_super)(struct supertype *st, struct mdinfo *info);
485
486 /* Check if the given metadata is flagged as belonging to "this"
487 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
488 */
489 int (*match_home)(struct supertype *st, char *homehost);
490
491 /* Make one of several generic modifications to metadata
492 * prior to assembly (or other times).
493 * sparc2.2 - first bug in early 0.90 metadata
494 * super-minor - change name of 0.90 metadata
495 * summaries - 'correct' any redundant data
496 * resync - mark array as dirty to trigger a resync.
497 * uuid - set new uuid - only 0.90 or 1.x
498 * name - change the name of the array (where supported)
499 * homehost - change which host this array is tied to.
500 * devicesize - If metadata is at start of device, change recorded
501 * device size to match actual device size
502 * byteorder - swap bytes for 0.90 metadata
503 *
504 * force-one - mark that device as uptodate, not old or failed.
505 * force-array - mark array as clean if it would not otherwise
506 * assemble
507 * assemble - not sure how this is different from force-one...
508 * linear-grow-new - add a new device to a linear array, but don't
509 * change the size: so superblock still matches
510 * linear-grow-update - now change the size of the array.
511 */
512 int (*update_super)(struct supertype *st, struct mdinfo *info,
513 char *update,
514 char *devname, int verbose,
515 int uuid_set, char *homehost);
516
517 /* Create new metadata for new array as described. This could
518 * be a new container, or an array in a pre-existing container.
519 * Also used to zero metadata prior to writing it to invalidate old
520 * metadata.
521 */
522 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
523 unsigned long long size, char *name,
524 char *homehost, int *uuid);
525
526 /* update the metadata to include new device, either at create or
527 * when hot-adding a spare.
528 */
529 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
530 int fd, char *devname);
531
532 /* Write metadata to one device when fixing problems or adding
533 * a new device.
534 */
535 int (*store_super)(struct supertype *st, int fd);
536
537 /* Write all metadata for this array.
538 */
539 int (*write_init_super)(struct supertype *st);
540 int (*compare_super)(struct supertype *st, struct supertype *tst);
541 int (*load_super)(struct supertype *st, int fd, char *devname);
542 struct supertype * (*match_metadata_desc)(char *arg);
543 __u64 (*avail_size)(struct supertype *st, __u64 size);
544 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
545 int delay, int write_behind,
546 unsigned long long size, int may_change, int major);
547 void (*locate_bitmap)(struct supertype *st, int fd);
548 int (*write_bitmap)(struct supertype *st, int fd);
549 void (*free_super)(struct supertype *st);
550
551 /* validate_geometry is called with an st returned by
552 * match_metadata_desc.
553 * It should check that the geometry described in compatible with
554 * the metadata type. It will be called repeatedly as devices
555 * added to validate changing size and new devices. If there are
556 * inter-device dependencies, it should record sufficient details
557 * so these can be validated.
558 * Both 'size' and '*freesize' are in sectors. chunk is bytes.
559 */
560 int (*validate_geometry)(struct supertype *st, int level, int layout,
561 int raiddisks,
562 int chunk, unsigned long long size,
563 char *subdev, unsigned long long *freesize,
564 int verbose);
565
566 struct mdinfo *(*container_content)(struct supertype *st);
567 /* Allow a metadata handler to override mdadm's default layouts */
568 int (*default_layout)(int level); /* optional */
569
570 /* for mdmon */
571 int (*open_new)(struct supertype *c, struct active_array *a,
572 char *inst);
573
574 /* Tell the metadata handler the current state of the array.
575 * This covers whether it is known to be consistent (no pending writes)
576 * and how far along a resync is known to have progressed
577 * (in a->resync_start).
578 * resync status is really irrelevant if the array is not consistent,
579 * but some metadata (DDF!) have a place to record the distinction.
580 * If 'consistent' is '2', then the array can mark it dirty if a
581 * resync/recovery/whatever is required, or leave it clean if not.
582 * Return value is 0 dirty (not consistent) and 1 if clean.
583 * it is only really important if consistent is passed in as '2'.
584 */
585 int (*set_array_state)(struct active_array *a, int consistent);
586
587 /* When the state of a device might have changed, we call set_disk to
588 * tell the metadata what the current state is.
589 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
590 * transitions.
591 * set_disk might be called when the state of the particular disk has
592 * not in fact changed.
593 */
594 void (*set_disk)(struct active_array *a, int n, int state);
595 void (*sync_metadata)(struct supertype *st);
596 void (*process_update)(struct supertype *st,
597 struct metadata_update *update);
598 void (*prepare_update)(struct supertype *st,
599 struct metadata_update *update);
600
601 /* activate_spare will check if the array is degraded and, if it
602 * is, try to find some spare space in the container.
603 * On success, it add appropriate updates (For process_update) to
604 * to the 'updates' list and returns a list of 'mdinfo' identifying
605 * the device, or devices as there might be multiple missing
606 * devices and multiple spares available.
607 */
608 struct mdinfo *(*activate_spare)(struct active_array *a,
609 struct metadata_update **updates);
610
611 int swapuuid; /* true if uuid is bigending rather than hostendian */
612 int external;
613 const char *name; /* canonical metadata name */
614 } super0, super1, super_ddf, *superlist[];
615
616 extern struct superswitch super_imsm;
617
618 struct metadata_update {
619 int len;
620 char *buf;
621 void *space; /* allocated space that monitor will use */
622 struct metadata_update *next;
623 };
624
625 /* A supertype holds a particular collection of metadata.
626 * It identifies the metadata type by the superswitch, and the particular
627 * sub-version of that metadata type.
628 * metadata read in or created is stored in 'sb' and 'info'.
629 * There are also fields used by mdmon to track containers.
630 *
631 * A supertype may refer to:
632 * Just an array, possibly in a container
633 * A container, not identifying any particular array
634 * Info read from just one device, not yet fully describing the array/container.
635 *
636 *
637 * A supertype is created by:
638 * super_by_fd
639 * guess_super
640 * dup_super
641 */
642 struct supertype {
643 struct superswitch *ss;
644 int minor_version;
645 int max_devs;
646 int container_dev; /* devnum of container */
647 char subarray[32]; /* name of array inside container */
648 void *sb;
649 void *info;
650 int loaded_container; /* Set if load_super found a container,
651 * not just one device */
652
653 struct metadata_update *updates;
654 struct metadata_update **update_tail;
655
656 /* extra stuff used by mdmon */
657 struct active_array *arrays;
658 int sock; /* listen to external programs */
659 int devnum;
660 char *devname; /* e.g. md0. This appears in metadata_verison:
661 * external:/md0/12
662 */
663 int devcnt;
664
665 struct mdinfo *devs;
666
667 };
668
669 extern struct supertype *super_by_fd(int fd);
670 extern struct supertype *guess_super(int fd);
671 extern struct supertype *dup_super(struct supertype *st);
672 extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
673 extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
674 mdu_disk_info_t *disk);
675 void wait_for(char *dev, int fd);
676
677 #if __GNUC__ < 3
678 struct stat64;
679 #endif
680
681 #define HAVE_NFTW we assume
682 #define HAVE_FTW
683
684 #ifdef __UCLIBC__
685 # include <features.h>
686 # ifndef __UCLIBC_HAS_LFS__
687 # define lseek64 lseek
688 # endif
689 # ifndef __UCLIBC_HAS_FTW__
690 # undef HAVE_FTW
691 # undef HAVE_NFTW
692 # endif
693 #endif
694
695 #ifdef __dietlibc__
696 # undef HAVE_NFTW
697 #endif
698
699 #if defined(__KLIBC__)
700 # undef HAVE_NFTW
701 # undef HAVE_FTW
702 #endif
703
704 #ifndef HAVE_NFTW
705 # define FTW_PHYS 1
706 # ifndef HAVE_FTW
707 struct FTW {};
708 # endif
709 #endif
710
711 #ifdef HAVE_FTW
712 # include <ftw.h>
713 #endif
714
715 extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
716
717
718 extern int Manage_ro(char *devname, int fd, int readonly);
719 extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
720 extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
721 extern int Manage_subdevs(char *devname, int fd,
722 mddev_dev_t devlist, int verbose);
723 extern int autodetect(void);
724 extern int Grow_Add_device(char *devname, int fd, char *newdev);
725 extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
726 extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
727 long long size,
728 int level, char *layout_str, int chunksize, int raid_disks);
729 extern int Grow_restart(struct supertype *st, struct mdinfo *info,
730 int *fdlist, int cnt, char *backup_file, int verbose);
731 extern int Grow_continue(int mdfd, struct supertype *st,
732 struct mdinfo *info, char *backup_file);
733
734 extern int Assemble(struct supertype *st, char *mddev,
735 mddev_ident_t ident,
736 mddev_dev_t devlist, char *backup_file,
737 int readonly, int runstop,
738 char *update, char *homehost, int require_homehost,
739 int verbose, int force);
740
741 extern int Build(char *mddev, int chunk, int level, int layout,
742 int raiddisks, mddev_dev_t devlist, int assume_clean,
743 char *bitmap_file, int bitmap_chunk, int write_behind,
744 int delay, int verbose, int autof, unsigned long long size);
745
746
747 extern int Create(struct supertype *st, char *mddev,
748 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
749 char *name, char *homehost, int *uuid,
750 int subdevs, mddev_dev_t devlist,
751 int runstop, int verbose, int force, int assume_clean,
752 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
753
754 extern int Detail(char *dev, int brief, int export, int test, char *homehost);
755 extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
756 extern int Query(char *dev);
757 extern int Examine(mddev_dev_t devlist, int brief, int export, int scan,
758 int SparcAdjust, struct supertype *forcest, char *homehost);
759 extern int Monitor(mddev_dev_t devlist,
760 char *mailaddr, char *alert_cmd,
761 int period, int daemonise, int scan, int oneshot,
762 int dosyslog, int test, char *pidfile, int increments);
763
764 extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
765 extern int Wait(char *dev);
766 extern int WaitClean(char *dev, int sock, int verbose);
767
768 extern int Incremental(char *devname, int verbose, int runstop,
769 struct supertype *st, char *homehost, int require_homehost,
770 int autof);
771 extern int Incremental_container(struct supertype *st, char *devname,
772 int verbose, int runstop, int autof,
773 int trustworthy);
774 extern void RebuildMap(void);
775 extern int IncrementalScan(int verbose);
776
777 extern int CreateBitmap(char *filename, int force, char uuid[16],
778 unsigned long chunksize, unsigned long daemon_sleep,
779 unsigned long write_behind,
780 unsigned long long array_size,
781 int major);
782 extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
783 extern int bitmap_update_uuid(int fd, int *uuid, int swap);
784 extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
785
786 extern int md_get_version(int fd);
787 extern int get_linux_version(void);
788 extern long long parse_size(char *size);
789 extern int parse_uuid(char *str, int uuid[4]);
790 extern int parse_layout_10(char *layout);
791 extern int parse_layout_faulty(char *layout);
792 extern int check_ext2(int fd, char *name);
793 extern int check_reiser(int fd, char *name);
794 extern int check_raid(int fd, char *name);
795
796 extern int get_mdp_major(void);
797 extern int dev_open(char *dev, int flags);
798 extern int open_dev(int devnum);
799 extern int open_dev_excl(int devnum);
800 extern int is_standard(char *dev, int *nump);
801 extern int same_dev(char *one, char *two);
802
803 extern int parse_auto(char *str, char *msg, int config);
804 extern mddev_ident_t conf_get_ident(char *dev);
805 extern mddev_dev_t conf_get_devs(void);
806 extern int conf_test_dev(char *devname);
807 extern int conf_test_metadata(const char *version);
808 extern struct createinfo *conf_get_create_info(void);
809 extern void set_conffile(char *file);
810 extern char *conf_get_mailaddr(void);
811 extern char *conf_get_mailfrom(void);
812 extern char *conf_get_program(void);
813 extern char *conf_get_homehost(int *require_homehostp);
814 extern char *conf_line(FILE *file);
815 extern char *conf_word(FILE *file, int allow_key);
816 extern int conf_name_is_free(char *name);
817 extern int devname_matches(char *name, char *match);
818 extern struct mddev_ident_s *conf_match(struct mdinfo *info, struct supertype *st);
819
820 extern void free_line(char *line);
821 extern int match_oneof(char *devices, char *devname);
822 extern void uuid_from_super(int uuid[4], mdp_super_t *super);
823 extern const int uuid_match_any[4];
824 extern int same_uuid(int a[4], int b[4], int swapuuid);
825 extern void copy_uuid(void *a, int b[4], int swapuuid);
826 extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
827 extern char *fname_from_uuid(struct supertype *st,
828 struct mdinfo *info, char *buf, char sep);
829 extern unsigned long calc_csum(void *super, int bytes);
830 extern int enough(int level, int raid_disks, int layout, int clean,
831 char *avail, int avail_disks);
832 extern int ask(char *mesg);
833 extern unsigned long long get_component_size(int fd);
834 extern void remove_partitions(int fd);
835 extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
836 int chunksize, unsigned long long devsize);
837 extern int flush_metadata_updates(struct supertype *st);
838 extern void append_metadata_update(struct supertype *st, void *buf, int len);
839 extern int assemble_container_content(struct supertype *st, int mdfd,
840 struct mdinfo *content, int runstop,
841 char *chosen_name, int verbose);
842
843 extern int add_disk(int mdfd, struct supertype *st,
844 struct mdinfo *sra, struct mdinfo *info);
845 extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
846
847 extern char *human_size(long long bytes);
848 extern char *human_size_brief(long long bytes);
849 extern void print_r10_layout(int layout);
850
851 #define NoMdDev (1<<23)
852 extern int find_free_devnum(int use_partitions);
853
854 extern void put_md_name(char *name);
855 extern char *get_md_name(int dev);
856
857 extern char DefaultConfFile[];
858
859 extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
860 char *chosen);
861 /* values for 'trustworthy' */
862 #define LOCAL 1
863 #define FOREIGN 2
864 #define METADATA 3
865 extern int open_mddev(char *dev, int report_errors);
866 extern int open_container(int fd);
867
868 extern int mdmon_running(int devnum);
869 extern int signal_mdmon(int devnum);
870 extern int check_env(char *name);
871 extern __u32 random32(void);
872 extern int start_mdmon(int devnum);
873
874 extern char *devnum2devname(int num);
875 extern int devname2devnum(char *name);
876 extern int stat2devnum(struct stat *st);
877 extern int fd2devnum(int fd);
878
879 static inline int dev2major(int d)
880 {
881 if (d >= 0)
882 return MD_MAJOR;
883 else
884 return get_mdp_major();
885 }
886
887 static inline int dev2minor(int d)
888 {
889 if (d >= 0)
890 return d;
891 return (-1-d) << MdpMinorShift;
892 }
893
894 static inline int ROUND_UP(int a, int base)
895 {
896 return ((a+base-1)/base)*base;
897 }
898
899 static inline int is_subarray(char *vers)
900 {
901 /* The version string for a 'subarray' (an array in a container)
902 * is
903 * /containername/componentname for normal read-write arrays
904 * -containername/componentname for read-only arrays.
905 * containername is e.g. md0, md_d1
906 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
907 */
908 return (*vers == '/' || *vers == '-');
909 }
910
911 #ifdef DEBUG
912 #define dprintf(fmt, arg...) \
913 fprintf(stderr, fmt, ##arg)
914 #else
915 #define dprintf(fmt, arg...) \
916 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
917 #endif
918 #include <assert.h>
919 #include <stdarg.h>
920 static inline int xasprintf(char **strp, const char *fmt, ...) {
921 va_list ap;
922 int ret;
923 va_start(ap, fmt);
924 ret = vasprintf(strp, fmt, ap);
925 va_end(ap);
926 assert(ret >= 0);
927 return ret;
928 }
929
930 #define LEVEL_MULTIPATH (-4)
931 #define LEVEL_LINEAR (-1)
932 #define LEVEL_FAULTY (-5)
933
934 /* kernel module doesn't know about these */
935 #define LEVEL_CONTAINER (-100)
936 #define LEVEL_UNSUPPORTED (-200)
937
938
939 /* faulty stuff */
940
941 #define WriteTransient 0
942 #define ReadTransient 1
943 #define WritePersistent 2
944 #define ReadPersistent 3
945 #define WriteAll 4 /* doesn't go to device */
946 #define ReadFixable 5
947 #define Modes 6
948
949 #define ClearErrors 31
950 #define ClearFaults 30
951
952 #define AllPersist 100 /* internal use only */
953 #define NoPersist 101
954
955 #define ModeMask 0x1f
956 #define ModeShift 5
957
958
959 #ifdef __TINYC__
960 #undef minor
961 #undef major
962 #undef makedev
963 #define minor(x) ((x)&0xff)
964 #define major(x) (((x)>>8)&0xff)
965 #define makedev(M,m) (((M)<<8) | (m))
966 #endif
967
968 /* for raid4/5/6 */
969 #define ALGORITHM_LEFT_ASYMMETRIC 0
970 #define ALGORITHM_RIGHT_ASYMMETRIC 1
971 #define ALGORITHM_LEFT_SYMMETRIC 2
972 #define ALGORITHM_RIGHT_SYMMETRIC 3
973
974 /* Define non-rotating (raid4) algorithms. These allow
975 * conversion of raid4 to raid5.
976 */
977 #define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
978 #define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
979
980 /* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
981 * Firstly, the exact positioning of the parity block is slightly
982 * different between the 'LEFT_*' modes of md and the "_N_*" modes
983 * of DDF.
984 * Secondly, or order of datablocks over which the Q syndrome is computed
985 * is different.
986 * Consequently we have different layouts for DDF/raid6 than md/raid6.
987 * These layouts are from the DDFv1.2 spec.
988 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
989 * leaves RLQ=3 as 'Vendor Specific'
990 */
991
992 #define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
993 #define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
994 #define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
995
996
997 /* For every RAID5 algorithm we define a RAID6 algorithm
998 * with exactly the same layout for data and parity, and
999 * with the Q block always on the last device (N-1).
1000 * This allows trivial conversion from RAID5 to RAID6
1001 */
1002 #define ALGORITHM_LEFT_ASYMMETRIC_6 16
1003 #define ALGORITHM_RIGHT_ASYMMETRIC_6 17
1004 #define ALGORITHM_LEFT_SYMMETRIC_6 18
1005 #define ALGORITHM_RIGHT_SYMMETRIC_6 19
1006 #define ALGORITHM_PARITY_0_6 20
1007 #define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1008