]> git.ipfire.org Git - thirdparty/mdadm.git/blob - mdadm.h
Fix required to enable RAID arrays on SAS disks.
[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 unsigned long long resync_start;
150 unsigned long safe_mode_delay; /* ms delay to mark clean */
151 int new_level, delta_disks, new_layout, new_chunk;
152 int errors;
153 int cache_size; /* size of raid456 stripe cache*/
154 int mismatch_cnt;
155 char text_version[50];
156 void *update_private; /* for passing metadata-format
157 * specific update data
158 * between successive calls to
159 * update_super()
160 */
161
162 int container_member; /* for assembling external-metatdata arrays
163 * This is to be used internally by metadata
164 * handler only */
165
166 char sys_name[20];
167 struct mdinfo *devs;
168 struct mdinfo *next;
169
170 /* Device info for mdmon: */
171 int state_fd;
172 #define DS_FAULTY 1
173 #define DS_INSYNC 2
174 #define DS_WRITE_MOSTLY 4
175 #define DS_SPARE 8
176 #define DS_BLOCKED 16
177 #define DS_REMOVE 1024
178 #define DS_UNBLOCK 2048
179 int prev_state, curr_state, next_state;
180
181 };
182
183 struct createinfo {
184 int uid;
185 int gid;
186 int autof;
187 int mode;
188 int symlinks;
189 struct supertype *supertype;
190 };
191
192 #define Name "mdadm"
193
194 enum mode {
195 ASSEMBLE=1,
196 BUILD,
197 CREATE,
198 MANAGE,
199 MISC,
200 MONITOR,
201 GROW,
202 INCREMENTAL,
203 AUTODETECT,
204 };
205
206 extern char short_options[];
207 extern char short_bitmap_options[];
208 extern char short_bitmap_auto_options[];
209 extern struct option long_options[];
210 extern char Version[], Usage[], Help[], OptionHelp[],
211 Help_create[], Help_build[], Help_assemble[], Help_grow[],
212 Help_incr[],
213 Help_manage[], Help_misc[], Help_monitor[], Help_config[];
214
215 /* for option that don't have short equivilents, we assign arbitrary
216 * small numbers. '1' means an undecorated option, so we start at '2'.
217 */
218 enum special_options {
219 AssumeClean = 2,
220 BitmapChunk,
221 WriteBehind,
222 ReAdd,
223 NoDegraded,
224 Sparc22,
225 BackupFile,
226 HomeHost,
227 AutoHomeHost,
228 Symlinks,
229 AutoDetect,
230 Waitclean,
231 DetailPlatform,
232 };
233
234 /* structures read from config file */
235 /* List of mddevice names and identifiers
236 * Identifiers can be:
237 * uuid=128-hex-uuid
238 * super-minor=decimal-minor-number-from-superblock
239 * devices=comma,separated,list,of,device,names,with,wildcards
240 *
241 * If multiple fields are present, the intersection of all matching
242 * devices is considered
243 */
244 #define UnSet (0xfffe)
245 typedef struct mddev_ident_s {
246 char *devname;
247
248 int uuid_set;
249 int uuid[4];
250 char name[33];
251
252 unsigned int super_minor;
253
254 char *devices; /* comma separated list of device
255 * names with wild cards
256 */
257 int level;
258 unsigned int raid_disks;
259 unsigned int spare_disks;
260 struct supertype *st;
261 int autof; /* 1 for normal, 2 for partitioned */
262 char *spare_group;
263 char *bitmap_file;
264 int bitmap_fd;
265
266 char *container; /* /dev/whatever name of container, or
267 * uuid of container. You would expect
268 * this to be the 'devname' or UUID
269 * of some other entry.
270 */
271 char *member; /* subarray within a container */
272
273 struct mddev_ident_s *next;
274 } *mddev_ident_t;
275
276 /* List of device names - wildcards expanded */
277 typedef struct mddev_dev_s {
278 char *devname;
279 char disposition; /* 'a' for add, 'r' for remove, 'f' for fail.
280 * Not set for names read from .config
281 */
282 char writemostly; /* 1 for 'set writemostly', 2 for 'clear writemostly' */
283 char re_add;
284 char used; /* set when used */
285 struct mdinfo *content; /* If devname is a container, this might list
286 * the remaining member arrays. */
287 struct mddev_dev_s *next;
288 } *mddev_dev_t;
289
290 typedef struct mapping {
291 char *name;
292 int num;
293 } mapping_t;
294
295
296 struct mdstat_ent {
297 char *dev;
298 int devnum;
299 int active;
300 char *level;
301 char *pattern; /* U or up, _ for down */
302 int percent; /* -1 if no resync */
303 int resync; /* 1 if resync, 0 if recovery */
304 int devcnt;
305 int raid_disks;
306 int chunk_size;
307 char * metadata_version;
308 struct mdstat_ent *next;
309 };
310
311 extern struct mdstat_ent *mdstat_read(int hold, int start);
312 extern void free_mdstat(struct mdstat_ent *ms);
313 extern void mdstat_wait(int seconds);
314 extern void mdstat_wait_fd(int fd, const sigset_t *sigmask);
315 extern int mddev_busy(int devnum);
316
317 struct map_ent {
318 struct map_ent *next;
319 int devnum;
320 char metadata[20];
321 int uuid[4];
322 int bad;
323 char *path;
324 };
325 extern int map_update(struct map_ent **mpp, int devnum, char *metadata,
326 int uuid[4], char *path);
327 extern struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4]);
328 extern struct map_ent *map_by_devnum(struct map_ent **map, int devnum);
329 extern struct map_ent *map_by_name(struct map_ent **map, char *name);
330 extern void map_read(struct map_ent **melp);
331 extern int map_write(struct map_ent *mel);
332 extern void map_delete(struct map_ent **mapp, int devnum);
333 extern void map_free(struct map_ent *map);
334 extern void map_add(struct map_ent **melp,
335 int devnum, char *metadata, int uuid[4], char *path);
336 extern int map_lock(struct map_ent **melp);
337 extern void map_unlock(struct map_ent **melp);
338
339 /* various details can be requested */
340 enum sysfs_read_flags {
341 GET_LEVEL = (1 << 0),
342 GET_LAYOUT = (1 << 1),
343 GET_COMPONENT = (1 << 2),
344 GET_CHUNK = (1 << 3),
345 GET_CACHE = (1 << 4),
346 GET_MISMATCH = (1 << 5),
347 GET_VERSION = (1 << 6),
348 GET_DISKS = (1 << 7),
349 GET_DEGRADED = (1 << 8),
350 GET_SAFEMODE = (1 << 9),
351 GET_DEVS = (1 << 10), /* gets role, major, minor */
352 GET_OFFSET = (1 << 11),
353 GET_SIZE = (1 << 12),
354 GET_STATE = (1 << 13),
355 GET_ERROR = (1 << 14),
356 SKIP_GONE_DEVS = (1 << 15),
357 };
358
359 /* If fd >= 0, get the array it is open on,
360 * else use devnum. >=0 -> major9. <0.....
361 */
362 extern int sysfs_open(int devnum, char *devname, char *attr);
363 extern void sysfs_init(struct mdinfo *mdi, int fd, int devnum);
364 extern void sysfs_free(struct mdinfo *sra);
365 extern struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options);
366 extern int sysfs_attr_match(const char *attr, const char *str);
367 extern int sysfs_match_word(const char *word, char **list);
368 extern int sysfs_set_str(struct mdinfo *sra, struct mdinfo *dev,
369 char *name, char *val);
370 extern int sysfs_set_num(struct mdinfo *sra, struct mdinfo *dev,
371 char *name, unsigned long long val);
372 extern int sysfs_uevent(struct mdinfo *sra, char *event);
373 extern int sysfs_get_fd(struct mdinfo *sra, struct mdinfo *dev,
374 char *name);
375 extern int sysfs_fd_get_ll(int fd, unsigned long long *val);
376 extern int sysfs_get_ll(struct mdinfo *sra, struct mdinfo *dev,
377 char *name, unsigned long long *val);
378 extern int sysfs_fd_get_str(int fd, char *val, int size);
379 extern int sysfs_get_str(struct mdinfo *sra, struct mdinfo *dev,
380 char *name, char *val, int size);
381 extern int sysfs_set_safemode(struct mdinfo *sra, unsigned long ms);
382 extern int sysfs_set_array(struct mdinfo *info, int vers);
383 extern int sysfs_add_disk(struct mdinfo *sra, struct mdinfo *sd,
384 int in_sync);
385 extern int sysfs_disk_to_scsi_id(int fd, __u32 *id);
386 extern int sysfs_unique_holder(int devnum, long rdev);
387 extern int load_sys(char *path, char *buf);
388
389
390 extern int save_stripes(int *source, unsigned long long *offsets,
391 int raid_disks, int chunk_size, int level, int layout,
392 int nwrites, int *dest,
393 unsigned long long start, unsigned long long length,
394 char *buf);
395 extern int restore_stripes(int *dest, unsigned long long *offsets,
396 int raid_disks, int chunk_size, int level, int layout,
397 int source, unsigned long long read_offset,
398 unsigned long long start, unsigned long long length);
399
400 #ifndef Sendmail
401 #define Sendmail "/usr/lib/sendmail -t"
402 #endif
403
404 #define SYSLOG_FACILITY LOG_DAEMON
405
406 extern char *map_num(mapping_t *map, int num);
407 extern int map_name(mapping_t *map, char *name);
408 extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
409
410 extern char *map_dev(int major, int minor, int create);
411
412 struct active_array;
413 struct metadata_update;
414
415 /* A superswitch provides entry point the a metadata handler.
416 *
417 * The super_switch primarily operates on some "metadata" that
418 * is accessed via the 'supertype'.
419 * This metadata has one of three possible sources.
420 * 1/ It is read from a single device. In this case it may not completely
421 * describe the array or arrays as some information might be on other
422 * devices.
423 * 2/ It is read from all devices in a container. In this case all
424 * information is present.
425 * 3/ It is created by ->init_super / ->add_to_super. In this case it will
426 * be complete once enough ->add_to_super calls have completed.
427 *
428 * When creating an array inside a container, the metadata will be
429 * formed by a combination of 2 and 3. The metadata or the array is read,
430 * then new information is added.
431 *
432 * The metadata must sometimes have a concept of a 'current' array
433 * and a 'current' device.
434 * The 'current' array is set by init_super to be the newly created array,
435 * or is set by super_by_fd when it finds it is looking at an array inside
436 * a container.
437 *
438 * The 'current' device is either the device that the metadata was read from
439 * in case 1, or the last device added by add_to_super in case 3.
440 * Case 2 does not identify a 'current' device.
441 */
442 extern struct superswitch {
443
444 /* Used to report details of metadata read from a component
445 * device. ->load_super has been called.
446 */
447 void (*examine_super)(struct supertype *st, char *homehost);
448 void (*brief_examine_super)(struct supertype *st, int verbose);
449 void (*brief_examine_subarrays)(struct supertype *st, int verbose);
450 void (*export_examine_super)(struct supertype *st);
451
452 /* Used to report details of an active array.
453 * ->load_super was possibly given a 'component' string.
454 */
455 void (*detail_super)(struct supertype *st, char *homehost);
456 void (*brief_detail_super)(struct supertype *st);
457 void (*export_detail_super)(struct supertype *st);
458
459 /* Optional: platform hardware / firmware details */
460 int (*detail_platform)(int verbose, int enumerate_only);
461
462 /* Used:
463 * to get uuid to storing in bitmap metadata
464 * and 'reshape' backup-data metadata
465 * To see if a device is being re-added to an array it was part of.
466 */
467 void (*uuid_from_super)(struct supertype *st, int uuid[4]);
468
469 /* Extract generic details from metadata. This could be details about
470 * the container, or about an individual array within the container.
471 * The determination is made either by:
472 * load_super being given a 'component' string.
473 * validate_geometry determining what to create.
474 * The info includes both array information and device information.
475 * The particular device should be:
476 * The last device added by add_to_super
477 * The device the metadata was loaded from by load_super
478 */
479 void (*getinfo_super)(struct supertype *st, struct mdinfo *info);
480
481 /* Check if the given metadata is flagged as belonging to "this"
482 * host. 0 for 'no', 1 for 'yes', -1 for "Don't record homehost"
483 */
484 int (*match_home)(struct supertype *st, char *homehost);
485
486 /* Make one of several generic modifications to metadata
487 * prior to assembly (or other times).
488 * sparc2.2 - first bug in early 0.90 metadata
489 * super-minor - change name of 0.90 metadata
490 * summaries - 'correct' any redundant data
491 * resync - mark array as dirty to trigger a resync.
492 * uuid - set new uuid - only 0.90 or 1.x
493 * name - change the name of the array (where supported)
494 * homehost - change which host this array is tied to.
495 * devicesize - If metadata is at start of device, change recorded
496 * device size to match actual device size
497 * byteorder - swap bytes for 0.90 metadata
498 *
499 * force-one - mark that device as uptodate, not old or failed.
500 * force-array - mark array as clean if it would not otherwise
501 * assemble
502 * assemble - not sure how this is different from force-one...
503 * linear-grow-new - add a new device to a linear array, but don't
504 * change the size: so superblock still matches
505 * linear-grow-update - now change the size of the array.
506 */
507 int (*update_super)(struct supertype *st, struct mdinfo *info,
508 char *update,
509 char *devname, int verbose,
510 int uuid_set, char *homehost);
511
512 /* Create new metadata for new array as described. This could
513 * be a new container, or an array in a pre-existing container.
514 * Also used to zero metadata prior to writing it to invalidate old
515 * metadata.
516 */
517 int (*init_super)(struct supertype *st, mdu_array_info_t *info,
518 unsigned long long size, char *name,
519 char *homehost, int *uuid);
520
521 /* update the metadata to include new device, either at create or
522 * when hot-adding a spare.
523 */
524 int (*add_to_super)(struct supertype *st, mdu_disk_info_t *dinfo,
525 int fd, char *devname);
526
527 /* Write metadata to one device when fixing problems or adding
528 * a new device.
529 */
530 int (*store_super)(struct supertype *st, int fd);
531
532 /* Write all metadata for this array.
533 */
534 int (*write_init_super)(struct supertype *st);
535 int (*compare_super)(struct supertype *st, struct supertype *tst);
536 int (*load_super)(struct supertype *st, int fd, char *devname);
537 struct supertype * (*match_metadata_desc)(char *arg);
538 __u64 (*avail_size)(struct supertype *st, __u64 size);
539 int (*add_internal_bitmap)(struct supertype *st, int *chunkp,
540 int delay, int write_behind,
541 unsigned long long size, int may_change, int major);
542 void (*locate_bitmap)(struct supertype *st, int fd);
543 int (*write_bitmap)(struct supertype *st, int fd);
544 void (*free_super)(struct supertype *st);
545
546 /* validate_geometry is called with an st returned by
547 * match_metadata_desc.
548 * It should check that the geometry described in compatible with
549 * the metadata type. It will be called repeatedly as devices
550 * added to validate changing size and new devices. If there are
551 * inter-device dependencies, it should record sufficient details
552 * so these can be validated.
553 * Both 'size' and '*freesize' are in sectors. chunk is bytes.
554 */
555 int (*validate_geometry)(struct supertype *st, int level, int layout,
556 int raiddisks,
557 int chunk, unsigned long long size,
558 char *subdev, unsigned long long *freesize,
559 int verbose);
560
561 struct mdinfo *(*container_content)(struct supertype *st);
562 /* Allow a metadata handler to override mdadm's default layouts */
563 int (*default_layout)(int level); /* optional */
564
565 /* for mdmon */
566 int (*open_new)(struct supertype *c, struct active_array *a,
567 char *inst);
568
569 /* Tell the metadata handler the current state of the array.
570 * This covers whether it is known to be consistent (no pending writes)
571 * and how far along a resync is known to have progressed
572 * (in a->resync_start).
573 * resync status is really irrelevant if the array is not consistent,
574 * but some metadata (DDF!) have a place to record the distinction.
575 * If 'consistent' is '2', then the array can mark it dirty if a
576 * resync/recovery/whatever is required, or leave it clean if not.
577 * Return value is 0 dirty (not consistent) and 1 if clean.
578 * it is only really important if consistent is passed in as '2'.
579 */
580 int (*set_array_state)(struct active_array *a, int consistent);
581
582 /* When the state of a device might have changed, we call set_disk to
583 * tell the metadata what the current state is.
584 * Typically this happens on spare->in_sync and (spare|in_sync)->faulty
585 * transitions.
586 * set_disk might be called when the state of the particular disk has
587 * not in fact changed.
588 */
589 void (*set_disk)(struct active_array *a, int n, int state);
590 void (*sync_metadata)(struct supertype *st);
591 void (*process_update)(struct supertype *st,
592 struct metadata_update *update);
593 void (*prepare_update)(struct supertype *st,
594 struct metadata_update *update);
595
596 /* activate_spare will check if the array is degraded and, if it
597 * is, try to find some spare space in the container.
598 * On success, it add appropriate updates (For process_update) to
599 * to the 'updates' list and returns a list of 'mdinfo' identifying
600 * the device, or devices as there might be multiple missing
601 * devices and multiple spares available.
602 */
603 struct mdinfo *(*activate_spare)(struct active_array *a,
604 struct metadata_update **updates);
605
606 int swapuuid; /* true if uuid is bigending rather than hostendian */
607 int external;
608 const char *name; /* canonical metadata name */
609 } super0, super1, super_ddf, *superlist[];
610
611 extern struct superswitch super_imsm;
612
613 struct metadata_update {
614 int len;
615 char *buf;
616 void *space; /* allocated space that monitor will use */
617 struct metadata_update *next;
618 };
619
620 /* A supertype holds a particular collection of metadata.
621 * It identifies the metadata type by the superswitch, and the particular
622 * sub-version of that metadata type.
623 * metadata read in or created is stored in 'sb' and 'info'.
624 * There are also fields used by mdmon to track containers.
625 *
626 * A supertype may refer to:
627 * Just an array, possibly in a container
628 * A container, not identifying any particular array
629 * Info read from just one device, not yet fully describing the array/container.
630 *
631 *
632 * A supertype is created by:
633 * super_by_fd
634 * guess_super
635 * dup_super
636 */
637 struct supertype {
638 struct superswitch *ss;
639 int minor_version;
640 int max_devs;
641 int container_dev; /* devnum of container */
642 char subarray[32]; /* name of array inside container */
643 void *sb;
644 void *info;
645 int loaded_container; /* Set if load_super found a container,
646 * not just one device */
647
648 struct metadata_update *updates;
649 struct metadata_update **update_tail;
650
651 /* extra stuff used by mdmon */
652 struct active_array *arrays;
653 int sock; /* listen to external programs */
654 int devnum;
655 char *devname; /* e.g. md0. This appears in metadata_verison:
656 * external:/md0/12
657 */
658 int devcnt;
659
660 struct mdinfo *devs;
661
662 };
663
664 extern struct supertype *super_by_fd(int fd);
665 extern struct supertype *guess_super(int fd);
666 extern struct supertype *dup_super(struct supertype *st);
667 extern int get_dev_size(int fd, char *dname, unsigned long long *sizep);
668 extern void get_one_disk(int mdfd, mdu_array_info_t *ainf,
669 mdu_disk_info_t *disk);
670 void wait_for(char *dev, int fd);
671
672 #if __GNUC__ < 3
673 struct stat64;
674 #endif
675
676 #define HAVE_NFTW we assume
677 #define HAVE_FTW
678
679 #ifdef __UCLIBC__
680 # include <features.h>
681 # ifndef __UCLIBC_HAS_LFS__
682 # define lseek64 lseek
683 # endif
684 # ifndef __UCLIBC_HAS_FTW__
685 # undef HAVE_FTW
686 # undef HAVE_NFTW
687 # endif
688 #endif
689
690 #ifdef __dietlibc__
691 # undef HAVE_NFTW
692 #endif
693
694 #if defined(__KLIBC__)
695 # undef HAVE_NFTW
696 # undef HAVE_FTW
697 #endif
698
699 #ifndef HAVE_NFTW
700 # define FTW_PHYS 1
701 # ifndef HAVE_FTW
702 struct FTW {};
703 # endif
704 #endif
705
706 #ifdef HAVE_FTW
707 # include <ftw.h>
708 #endif
709
710 extern int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s);
711
712
713 extern int Manage_ro(char *devname, int fd, int readonly);
714 extern int Manage_runstop(char *devname, int fd, int runstop, int quiet);
715 extern int Manage_resize(char *devname, int fd, long long size, int raid_disks);
716 extern int Manage_subdevs(char *devname, int fd,
717 mddev_dev_t devlist, int verbose);
718 extern int autodetect(void);
719 extern int Grow_Add_device(char *devname, int fd, char *newdev);
720 extern int Grow_addbitmap(char *devname, int fd, char *file, int chunk, int delay, int write_behind, int force);
721 extern int Grow_reshape(char *devname, int fd, int quiet, char *backup_file,
722 long long size,
723 int level, char *layout_str, int chunksize, int raid_disks);
724 extern int Grow_restart(struct supertype *st, struct mdinfo *info,
725 int *fdlist, int cnt, char *backup_file, int verbose);
726 extern int Grow_continue(int mdfd, struct supertype *st,
727 struct mdinfo *info, char *backup_file);
728
729 extern int Assemble(struct supertype *st, char *mddev,
730 mddev_ident_t ident,
731 mddev_dev_t devlist, char *backup_file,
732 int readonly, int runstop,
733 char *update, char *homehost, int require_homehost,
734 int verbose, int force);
735
736 extern int Build(char *mddev, int chunk, int level, int layout,
737 int raiddisks, mddev_dev_t devlist, int assume_clean,
738 char *bitmap_file, int bitmap_chunk, int write_behind,
739 int delay, int verbose, int autof, unsigned long long size);
740
741
742 extern int Create(struct supertype *st, char *mddev,
743 int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks,
744 char *name, char *homehost, int *uuid,
745 int subdevs, mddev_dev_t devlist,
746 int runstop, int verbose, int force, int assume_clean,
747 char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof);
748
749 extern int Detail(char *dev, int brief, int export, int test, char *homehost);
750 extern int Detail_Platform(struct superswitch *ss, int scan, int verbose);
751 extern int Query(char *dev);
752 extern int Examine(mddev_dev_t devlist, int brief, int export, int scan,
753 int SparcAdjust, struct supertype *forcest, char *homehost);
754 extern int Monitor(mddev_dev_t devlist,
755 char *mailaddr, char *alert_cmd,
756 int period, int daemonise, int scan, int oneshot,
757 int dosyslog, int test, char *pidfile, int increments);
758
759 extern int Kill(char *dev, struct supertype *st, int force, int quiet, int noexcl);
760 extern int Wait(char *dev);
761 extern int WaitClean(char *dev, int sock, int verbose);
762
763 extern int Incremental(char *devname, int verbose, int runstop,
764 struct supertype *st, char *homehost, int require_homehost,
765 int autof);
766 extern int Incremental_container(struct supertype *st, char *devname,
767 int verbose, int runstop, int autof,
768 int trustworthy);
769 extern void RebuildMap(void);
770 extern int IncrementalScan(int verbose);
771
772 extern int CreateBitmap(char *filename, int force, char uuid[16],
773 unsigned long chunksize, unsigned long daemon_sleep,
774 unsigned long write_behind,
775 unsigned long long array_size,
776 int major);
777 extern int ExamineBitmap(char *filename, int brief, struct supertype *st);
778 extern int bitmap_update_uuid(int fd, int *uuid, int swap);
779 extern unsigned long bitmap_sectors(struct bitmap_super_s *bsb);
780
781 extern int md_get_version(int fd);
782 extern int get_linux_version(void);
783 extern long long parse_size(char *size);
784 extern int parse_uuid(char *str, int uuid[4]);
785 extern int parse_layout_10(char *layout);
786 extern int parse_layout_faulty(char *layout);
787 extern int check_ext2(int fd, char *name);
788 extern int check_reiser(int fd, char *name);
789 extern int check_raid(int fd, char *name);
790
791 extern int get_mdp_major(void);
792 extern int dev_open(char *dev, int flags);
793 extern int open_dev(int devnum);
794 extern int open_dev_excl(int devnum);
795 extern int is_standard(char *dev, int *nump);
796 extern int same_dev(char *one, char *two);
797
798 extern int parse_auto(char *str, char *msg, int config);
799 extern mddev_ident_t conf_get_ident(char *dev);
800 extern mddev_dev_t conf_get_devs(void);
801 extern int conf_test_dev(char *devname);
802 extern int conf_test_metadata(const char *version);
803 extern struct createinfo *conf_get_create_info(void);
804 extern void set_conffile(char *file);
805 extern char *conf_get_mailaddr(void);
806 extern char *conf_get_mailfrom(void);
807 extern char *conf_get_program(void);
808 extern char *conf_get_homehost(int *require_homehostp);
809 extern char *conf_line(FILE *file);
810 extern char *conf_word(FILE *file, int allow_key);
811 extern int conf_name_is_free(char *name);
812 extern int devname_matches(char *name, char *match);
813 extern struct mddev_ident_s *conf_match(struct mdinfo *info, struct supertype *st);
814
815 extern void free_line(char *line);
816 extern int match_oneof(char *devices, char *devname);
817 extern void uuid_from_super(int uuid[4], mdp_super_t *super);
818 extern const int uuid_match_any[4];
819 extern int same_uuid(int a[4], int b[4], int swapuuid);
820 extern void copy_uuid(void *a, int b[4], int swapuuid);
821 extern char *__fname_from_uuid(int id[4], int swap, char *buf, char sep);
822 extern char *fname_from_uuid(struct supertype *st,
823 struct mdinfo *info, char *buf, char sep);
824 extern unsigned long calc_csum(void *super, int bytes);
825 extern int enough(int level, int raid_disks, int layout, int clean,
826 char *avail, int avail_disks);
827 extern int ask(char *mesg);
828 extern unsigned long long get_component_size(int fd);
829 extern void remove_partitions(int fd);
830 extern unsigned long long calc_array_size(int level, int raid_disks, int layout,
831 int chunksize, unsigned long long devsize);
832 extern int flush_metadata_updates(struct supertype *st);
833 extern void append_metadata_update(struct supertype *st, void *buf, int len);
834 extern int assemble_container_content(struct supertype *st, int mdfd,
835 struct mdinfo *content, int runstop,
836 char *chosen_name, int verbose);
837
838 extern int add_disk(int mdfd, struct supertype *st,
839 struct mdinfo *sra, struct mdinfo *info);
840 extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
841
842 extern char *human_size(long long bytes);
843 extern char *human_size_brief(long long bytes);
844 extern void print_r10_layout(int layout);
845
846 #define NoMdDev (1<<23)
847 extern int find_free_devnum(int use_partitions);
848
849 extern void put_md_name(char *name);
850 extern char *get_md_name(int dev);
851
852 extern char DefaultConfFile[];
853
854 extern int create_mddev(char *dev, char *name, int autof, int trustworthy,
855 char *chosen);
856 /* values for 'trustworthy' */
857 #define LOCAL 1
858 #define FOREIGN 2
859 #define METADATA 3
860 extern int open_mddev(char *dev, int report_errors);
861 extern int open_container(int fd);
862
863 extern int mdmon_running(int devnum);
864 extern int signal_mdmon(int devnum);
865 extern int check_env(char *name);
866 extern __u32 random32(void);
867 extern int start_mdmon(int devnum);
868
869 extern char *devnum2devname(int num);
870 extern int devname2devnum(char *name);
871 extern int stat2devnum(struct stat *st);
872 extern int fd2devnum(int fd);
873
874 static inline int dev2major(int d)
875 {
876 if (d >= 0)
877 return MD_MAJOR;
878 else
879 return get_mdp_major();
880 }
881
882 static inline int dev2minor(int d)
883 {
884 if (d >= 0)
885 return d;
886 return (-1-d) << MdpMinorShift;
887 }
888
889 static inline int ROUND_UP(int a, int base)
890 {
891 return ((a+base-1)/base)*base;
892 }
893
894 static inline int is_subarray(char *vers)
895 {
896 /* The version string for a 'subarray' (an array in a container)
897 * is
898 * /containername/componentname for normal read-write arrays
899 * -containername/componentname for read-only arrays.
900 * containername is e.g. md0, md_d1
901 * componentname is dependant on the metadata. e.g. '1' 'S1' ...
902 */
903 return (*vers == '/' || *vers == '-');
904 }
905
906 #ifdef DEBUG
907 #define dprintf(fmt, arg...) \
908 fprintf(stderr, fmt, ##arg)
909 #else
910 #define dprintf(fmt, arg...) \
911 ({ if (0) fprintf(stderr, fmt, ##arg); 0; })
912 #endif
913 #include <assert.h>
914 #include <stdarg.h>
915 static inline int xasprintf(char **strp, const char *fmt, ...) {
916 va_list ap;
917 int ret;
918 va_start(ap, fmt);
919 ret = vasprintf(strp, fmt, ap);
920 va_end(ap);
921 assert(ret >= 0);
922 return ret;
923 }
924
925 #define LEVEL_MULTIPATH (-4)
926 #define LEVEL_LINEAR (-1)
927 #define LEVEL_FAULTY (-5)
928
929 /* kernel module doesn't know about these */
930 #define LEVEL_CONTAINER (-100)
931 #define LEVEL_UNSUPPORTED (-200)
932
933
934 /* faulty stuff */
935
936 #define WriteTransient 0
937 #define ReadTransient 1
938 #define WritePersistent 2
939 #define ReadPersistent 3
940 #define WriteAll 4 /* doesn't go to device */
941 #define ReadFixable 5
942 #define Modes 6
943
944 #define ClearErrors 31
945 #define ClearFaults 30
946
947 #define AllPersist 100 /* internal use only */
948 #define NoPersist 101
949
950 #define ModeMask 0x1f
951 #define ModeShift 5
952
953
954 #ifdef __TINYC__
955 #undef minor
956 #undef major
957 #undef makedev
958 #define minor(x) ((x)&0xff)
959 #define major(x) (((x)>>8)&0xff)
960 #define makedev(M,m) (((M)<<8) | (m))
961 #endif
962
963 /* for raid4/5/6 */
964 #define ALGORITHM_LEFT_ASYMMETRIC 0
965 #define ALGORITHM_RIGHT_ASYMMETRIC 1
966 #define ALGORITHM_LEFT_SYMMETRIC 2
967 #define ALGORITHM_RIGHT_SYMMETRIC 3
968
969 /* Define non-rotating (raid4) algorithms. These allow
970 * conversion of raid4 to raid5.
971 */
972 #define ALGORITHM_PARITY_0 4 /* P or P,Q are initial devices */
973 #define ALGORITHM_PARITY_N 5 /* P or P,Q are final devices. */
974
975 /* DDF RAID6 layouts differ from md/raid6 layouts in two ways.
976 * Firstly, the exact positioning of the parity block is slightly
977 * different between the 'LEFT_*' modes of md and the "_N_*" modes
978 * of DDF.
979 * Secondly, or order of datablocks over which the Q syndrome is computed
980 * is different.
981 * Consequently we have different layouts for DDF/raid6 than md/raid6.
982 * These layouts are from the DDFv1.2 spec.
983 * Interestingly DDFv1.2-Errata-A does not specify N_CONTINUE but
984 * leaves RLQ=3 as 'Vendor Specific'
985 */
986
987 #define ALGORITHM_ROTATING_ZERO_RESTART 8 /* DDF PRL=6 RLQ=1 */
988 #define ALGORITHM_ROTATING_N_RESTART 9 /* DDF PRL=6 RLQ=2 */
989 #define ALGORITHM_ROTATING_N_CONTINUE 10 /*DDF PRL=6 RLQ=3 */
990
991
992 /* For every RAID5 algorithm we define a RAID6 algorithm
993 * with exactly the same layout for data and parity, and
994 * with the Q block always on the last device (N-1).
995 * This allows trivial conversion from RAID5 to RAID6
996 */
997 #define ALGORITHM_LEFT_ASYMMETRIC_6 16
998 #define ALGORITHM_RIGHT_ASYMMETRIC_6 17
999 #define ALGORITHM_LEFT_SYMMETRIC_6 18
1000 #define ALGORITHM_RIGHT_SYMMETRIC_6 19
1001 #define ALGORITHM_PARITY_0_6 20
1002 #define ALGORITHM_PARITY_N_6 ALGORITHM_PARITY_N
1003
1004 /* Define PATH_MAX in case we don't use glibc or standard library does
1005 * not have PATH_MAX defined. Assume max path length is 4K characters.
1006 */
1007 #ifndef PATH_MAX
1008 #define PATH_MAX 4096
1009 #endif
1010