2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2010 Alexander Motin <mav@FreeBSD.org>
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 #include <sys/param.h>
39 #include <sys/sysctl.h>
42 #define G_RAID_CLASS_NAME "RAID"
44 #define G_RAID_MAGIC "GEOM::RAID"
46 #define G_RAID_VERSION 0
48 struct g_raid_md_object;
49 struct g_raid_tr_object;
51 #define G_RAID_DEVICE_FLAG_NOAUTOSYNC 0x0000000000000001ULL
52 #define G_RAID_DEVICE_FLAG_NOFAILSYNC 0x0000000000000002ULL
53 #define G_RAID_DEVICE_FLAG_MASK (G_RAID_DEVICE_FLAG_NOAUTOSYNC | \
54 G_RAID_DEVICE_FLAG_NOFAILSYNC)
57 extern u_int g_raid_aggressive_spare;
58 extern u_int g_raid_debug;
59 extern int g_raid_enable;
60 extern int g_raid_read_err_thresh;
61 extern u_int g_raid_start_timeout;
62 extern struct g_class g_raid_class;
64 #define G_RAID_DEBUG(lvl, fmt, ...) do { \
65 if (g_raid_debug >= (lvl)) { \
66 if (g_raid_debug > 0) { \
67 printf("GEOM_RAID[%u]: " fmt "\n", \
68 lvl, ## __VA_ARGS__); \
70 printf("GEOM_RAID: " fmt "\n", \
75 #define G_RAID_DEBUG1(lvl, sc, fmt, ...) do { \
76 if (g_raid_debug >= (lvl)) { \
77 if (g_raid_debug > 0) { \
78 printf("GEOM_RAID[%u]: %s: " fmt "\n", \
79 lvl, (sc)->sc_name, ## __VA_ARGS__); \
81 printf("GEOM_RAID: %s: " fmt "\n", \
82 (sc)->sc_name, ## __VA_ARGS__); \
86 #define G_RAID_LOGREQ(lvl, bp, fmt, ...) do { \
87 if (g_raid_debug >= (lvl)) { \
88 if (g_raid_debug > 0) { \
89 printf("GEOM_RAID[%u]: " fmt " ", \
90 lvl, ## __VA_ARGS__); \
92 printf("GEOM_RAID: " fmt " ", ## __VA_ARGS__); \
99 * Flags we use to distinguish I/O initiated by the TR layer to maintain
100 * the volume's characteristics, fix subdisks, extra copies of data, etc.
102 * G_RAID_BIO_FLAG_SYNC I/O to update an extra copy of the data
103 * for RAID volumes that maintain extra data
104 * and need to rebuild that data.
105 * G_RAID_BIO_FLAG_REMAP I/O done to try to provoke a subdisk into
106 * doing some desirable action such as bad
107 * block remapping after we detect a bad part
109 * G_RAID_BIO_FLAG_LOCKED I/O holds range lock that should re released.
111 * and the following meta item:
112 * G_RAID_BIO_FLAG_SPECIAL And of the I/O flags that need to make it
113 * through the range locking which would
114 * otherwise defer the I/O until after that
117 #define G_RAID_BIO_FLAG_SYNC 0x01
118 #define G_RAID_BIO_FLAG_REMAP 0x02
119 #define G_RAID_BIO_FLAG_SPECIAL \
120 (G_RAID_BIO_FLAG_SYNC|G_RAID_BIO_FLAG_REMAP)
121 #define G_RAID_BIO_FLAG_LOCKED 0x80
126 void *l_callback_arg;
128 LIST_ENTRY(g_raid_lock) l_next;
131 #define G_RAID_EVENT_WAIT 0x01
132 #define G_RAID_EVENT_VOLUME 0x02
133 #define G_RAID_EVENT_SUBDISK 0x04
134 #define G_RAID_EVENT_DISK 0x08
135 #define G_RAID_EVENT_DONE 0x10
136 struct g_raid_event {
141 TAILQ_ENTRY(g_raid_event) e_next;
143 #define G_RAID_DISK_S_NONE 0x00 /* State is unknown. */
144 #define G_RAID_DISK_S_OFFLINE 0x01 /* Missing disk placeholder. */
145 #define G_RAID_DISK_S_DISABLED 0x02 /* Disabled. */
146 #define G_RAID_DISK_S_FAILED 0x03 /* Failed. */
147 #define G_RAID_DISK_S_STALE_FAILED 0x04 /* Old failed. */
148 #define G_RAID_DISK_S_SPARE 0x05 /* Hot-spare. */
149 #define G_RAID_DISK_S_STALE 0x06 /* Old disk, unused now. */
150 #define G_RAID_DISK_S_ACTIVE 0x07 /* Operational. */
152 #define G_RAID_DISK_E_DISCONNECTED 0x01
155 struct g_raid_softc *d_softc; /* Back-pointer to softc. */
156 struct g_consumer *d_consumer; /* GEOM disk consumer. */
157 void *d_md_data; /* Disk's metadata storage. */
158 struct g_kerneldump d_kd; /* Kernel dumping method/args. */
159 int d_candelete; /* BIO_DELETE supported. */
160 uint64_t d_flags; /* Additional flags. */
161 u_int d_state; /* Disk state. */
162 u_int d_load; /* Disk average load. */
163 off_t d_last_offset; /* Last head offset. */
164 int d_read_errs; /* Count of the read errors */
165 TAILQ_HEAD(, g_raid_subdisk) d_subdisks; /* List of subdisks. */
166 TAILQ_ENTRY(g_raid_disk) d_next; /* Next disk in the node. */
169 #define G_RAID_SUBDISK_S_NONE 0x00 /* Absent. */
170 #define G_RAID_SUBDISK_S_FAILED 0x01 /* Failed. */
171 #define G_RAID_SUBDISK_S_NEW 0x02 /* Blank. */
172 #define G_RAID_SUBDISK_S_REBUILD 0x03 /* Blank + rebuild. */
173 #define G_RAID_SUBDISK_S_UNINITIALIZED 0x04 /* Disk of the new volume. */
174 #define G_RAID_SUBDISK_S_STALE 0x05 /* Dirty. */
175 #define G_RAID_SUBDISK_S_RESYNC 0x06 /* Dirty + check/repair. */
176 #define G_RAID_SUBDISK_S_ACTIVE 0x07 /* Usable. */
178 #define G_RAID_SUBDISK_E_NEW 0x01 /* A new subdisk has arrived */
179 #define G_RAID_SUBDISK_E_FAILED 0x02 /* A subdisk failed, but remains in volume */
180 #define G_RAID_SUBDISK_E_DISCONNECTED 0x03 /* A subdisk removed from volume. */
181 #define G_RAID_SUBDISK_E_FIRST_TR_PRIVATE 0x80 /* translation private events */
183 #define G_RAID_SUBDISK_POS(sd) \
184 ((sd)->sd_disk ? ((sd)->sd_disk->d_last_offset - (sd)->sd_offset) : 0)
185 #define G_RAID_SUBDISK_TRACK_SIZE (1 * 1024 * 1024)
186 #define G_RAID_SUBDISK_LOAD(sd) \
187 ((sd)->sd_disk ? ((sd)->sd_disk->d_load) : 0)
188 #define G_RAID_SUBDISK_LOAD_SCALE 256
190 struct g_raid_subdisk {
191 struct g_raid_softc *sd_softc; /* Back-pointer to softc. */
192 struct g_raid_disk *sd_disk; /* Where this subdisk lives. */
193 struct g_raid_volume *sd_volume; /* Volume, sd is a part of. */
194 off_t sd_offset; /* Offset on the disk. */
195 off_t sd_size; /* Size on the disk. */
196 u_int sd_pos; /* Position in volume. */
197 u_int sd_state; /* Subdisk state. */
198 off_t sd_rebuild_pos; /* Rebuild position. */
199 int sd_recovery; /* Count of recovery reqs. */
200 TAILQ_ENTRY(g_raid_subdisk) sd_next; /* Next subdisk on disk. */
203 #define G_RAID_MAX_SUBDISKS 16
204 #define G_RAID_MAX_VOLUMENAME 32
206 #define G_RAID_VOLUME_S_STARTING 0x00
207 #define G_RAID_VOLUME_S_BROKEN 0x01
208 #define G_RAID_VOLUME_S_DEGRADED 0x02
209 #define G_RAID_VOLUME_S_SUBOPTIMAL 0x03
210 #define G_RAID_VOLUME_S_OPTIMAL 0x04
211 #define G_RAID_VOLUME_S_UNSUPPORTED 0x05
212 #define G_RAID_VOLUME_S_STOPPED 0x06
214 #define G_RAID_VOLUME_S_ALIVE(s) \
215 ((s) == G_RAID_VOLUME_S_DEGRADED || \
216 (s) == G_RAID_VOLUME_S_SUBOPTIMAL || \
217 (s) == G_RAID_VOLUME_S_OPTIMAL)
219 #define G_RAID_VOLUME_E_DOWN 0x00
220 #define G_RAID_VOLUME_E_UP 0x01
221 #define G_RAID_VOLUME_E_START 0x10
222 #define G_RAID_VOLUME_E_STARTMD 0x11
224 #define G_RAID_VOLUME_RL_RAID0 0x00
225 #define G_RAID_VOLUME_RL_RAID1 0x01
226 #define G_RAID_VOLUME_RL_RAID3 0x03
227 #define G_RAID_VOLUME_RL_RAID4 0x04
228 #define G_RAID_VOLUME_RL_RAID5 0x05
229 #define G_RAID_VOLUME_RL_RAID6 0x06
230 #define G_RAID_VOLUME_RL_RAIDMDF 0x07
231 #define G_RAID_VOLUME_RL_RAID1E 0x11
232 #define G_RAID_VOLUME_RL_SINGLE 0x0f
233 #define G_RAID_VOLUME_RL_CONCAT 0x1f
234 #define G_RAID_VOLUME_RL_RAID5E 0x15
235 #define G_RAID_VOLUME_RL_RAID5EE 0x25
236 #define G_RAID_VOLUME_RL_RAID5R 0x35
237 #define G_RAID_VOLUME_RL_UNKNOWN 0xff
239 #define G_RAID_VOLUME_RLQ_NONE 0x00
240 #define G_RAID_VOLUME_RLQ_R1SM 0x00
241 #define G_RAID_VOLUME_RLQ_R1MM 0x01
242 #define G_RAID_VOLUME_RLQ_R3P0 0x00
243 #define G_RAID_VOLUME_RLQ_R3PN 0x01
244 #define G_RAID_VOLUME_RLQ_R4P0 0x00
245 #define G_RAID_VOLUME_RLQ_R4PN 0x01
246 #define G_RAID_VOLUME_RLQ_R5RA 0x00
247 #define G_RAID_VOLUME_RLQ_R5RS 0x01
248 #define G_RAID_VOLUME_RLQ_R5LA 0x02
249 #define G_RAID_VOLUME_RLQ_R5LS 0x03
250 #define G_RAID_VOLUME_RLQ_R6RA 0x00
251 #define G_RAID_VOLUME_RLQ_R6RS 0x01
252 #define G_RAID_VOLUME_RLQ_R6LA 0x02
253 #define G_RAID_VOLUME_RLQ_R6LS 0x03
254 #define G_RAID_VOLUME_RLQ_RMDFRA 0x00
255 #define G_RAID_VOLUME_RLQ_RMDFRS 0x01
256 #define G_RAID_VOLUME_RLQ_RMDFLA 0x02
257 #define G_RAID_VOLUME_RLQ_RMDFLS 0x03
258 #define G_RAID_VOLUME_RLQ_R1EA 0x00
259 #define G_RAID_VOLUME_RLQ_R1EO 0x01
260 #define G_RAID_VOLUME_RLQ_R5ERA 0x00
261 #define G_RAID_VOLUME_RLQ_R5ERS 0x01
262 #define G_RAID_VOLUME_RLQ_R5ELA 0x02
263 #define G_RAID_VOLUME_RLQ_R5ELS 0x03
264 #define G_RAID_VOLUME_RLQ_R5EERA 0x00
265 #define G_RAID_VOLUME_RLQ_R5EERS 0x01
266 #define G_RAID_VOLUME_RLQ_R5EELA 0x02
267 #define G_RAID_VOLUME_RLQ_R5EELS 0x03
268 #define G_RAID_VOLUME_RLQ_R5RRA 0x00
269 #define G_RAID_VOLUME_RLQ_R5RRS 0x01
270 #define G_RAID_VOLUME_RLQ_R5RLA 0x02
271 #define G_RAID_VOLUME_RLQ_R5RLS 0x03
272 #define G_RAID_VOLUME_RLQ_UNKNOWN 0xff
274 struct g_raid_volume;
276 struct g_raid_volume {
277 struct g_raid_softc *v_softc; /* Back-pointer to softc. */
278 struct g_provider *v_provider; /* GEOM provider. */
279 struct g_raid_subdisk v_subdisks[G_RAID_MAX_SUBDISKS];
280 /* Subdisks of this volume. */
281 void *v_md_data; /* Volume's metadata storage. */
282 struct g_raid_tr_object *v_tr; /* Transformation object. */
283 char v_name[G_RAID_MAX_VOLUMENAME];
285 u_int v_state; /* Volume state. */
286 u_int v_raid_level; /* Array RAID level. */
287 u_int v_raid_level_qualifier; /* RAID level det. */
288 u_int v_disks_count; /* Number of disks in array. */
289 u_int v_mdf_pdisks; /* Number of parity disks
291 uint16_t v_mdf_polynomial; /* Polynomial for RAIDMDF. */
292 uint8_t v_mdf_method; /* Generation method for RAIDMDF. */
293 u_int v_strip_size; /* Array strip size. */
294 u_int v_rotate_parity; /* Rotate RAID5R parity
295 after numer of stripes. */
296 u_int v_sectorsize; /* Volume sector size. */
297 off_t v_mediasize; /* Volume media size. */
298 struct bio_queue_head v_inflight; /* In-flight write requests. */
299 struct bio_queue_head v_locked; /* Blocked I/O requests. */
300 LIST_HEAD(, g_raid_lock) v_locks; /* List of locked regions. */
301 int v_pending_lock; /* writes to locked region */
302 int v_dirty; /* Volume is DIRTY. */
303 struct timeval v_last_done; /* Time of the last I/O. */
304 time_t v_last_write; /* Time of the last write. */
305 u_int v_writes; /* Number of active writes. */
306 struct root_hold_token *v_rootmount; /* Root mount delay token. */
307 int v_starting; /* Volume is starting */
308 int v_stopping; /* Volume is stopping */
309 int v_provider_open; /* Number of opens. */
310 int v_global_id; /* Global volume ID (rX). */
311 int v_read_only; /* Volume is read-only. */
312 TAILQ_ENTRY(g_raid_volume) v_next; /* List of volumes entry. */
313 LIST_ENTRY(g_raid_volume) v_global_next; /* Global list entry. */
316 #define G_RAID_NODE_E_WAKE 0x00
317 #define G_RAID_NODE_E_START 0x01
319 struct g_raid_softc {
320 struct g_raid_md_object *sc_md; /* Metadata object. */
321 struct g_geom *sc_geom; /* GEOM class instance. */
322 uint64_t sc_flags; /* Additional flags. */
323 TAILQ_HEAD(, g_raid_volume) sc_volumes; /* List of volumes. */
324 TAILQ_HEAD(, g_raid_disk) sc_disks; /* List of disks. */
325 struct sx sc_lock; /* Main node lock. */
326 struct proc *sc_worker; /* Worker process. */
327 struct mtx sc_queue_mtx; /* Worker queues lock. */
328 TAILQ_HEAD(, g_raid_event) sc_events; /* Worker events queue. */
329 struct bio_queue_head sc_queue; /* Worker I/O queue. */
330 int sc_stopping; /* Node is stopping */
332 #define sc_name sc_geom->name
334 SYSCTL_DECL(_kern_geom_raid);
337 * KOBJ parent class of metadata processing modules.
339 struct g_raid_md_class {
343 LIST_ENTRY(g_raid_md_class) mdc_list;
347 * KOBJ instance of metadata processing module.
349 struct g_raid_md_object {
351 struct g_raid_md_class *mdo_class;
352 struct g_raid_softc *mdo_softc; /* Back-pointer to softc. */
355 int g_raid_md_modevent(module_t, int, void *);
357 #define G_RAID_MD_DECLARE(name, label) \
358 static moduledata_t g_raid_md_##name##_mod = { \
359 "g_raid_md_" __XSTRING(name), \
360 g_raid_md_modevent, \
361 &g_raid_md_##name##_class \
363 DECLARE_MODULE(g_raid_md_##name, g_raid_md_##name##_mod, \
364 SI_SUB_DRIVERS, SI_ORDER_SECOND); \
365 MODULE_DEPEND(g_raid_md_##name, geom_raid, 0, 0, 0); \
366 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, CTLFLAG_RD, \
367 NULL, label " metadata module"); \
368 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
369 CTLFLAG_RWTUN, &g_raid_md_##name##_class.mdc_enable, 0, \
370 "Enable " label " metadata format taste")
373 * KOBJ parent class of data transformation modules.
375 struct g_raid_tr_class {
379 int trc_accept_unmapped;
380 LIST_ENTRY(g_raid_tr_class) trc_list;
384 * KOBJ instance of data transformation module.
386 struct g_raid_tr_object {
388 struct g_raid_tr_class *tro_class;
389 struct g_raid_volume *tro_volume; /* Back-pointer to volume. */
392 int g_raid_tr_modevent(module_t, int, void *);
394 #define G_RAID_TR_DECLARE(name, label) \
395 static moduledata_t g_raid_tr_##name##_mod = { \
396 "g_raid_tr_" __XSTRING(name), \
397 g_raid_tr_modevent, \
398 &g_raid_tr_##name##_class \
400 DECLARE_MODULE(g_raid_tr_##name, g_raid_tr_##name##_mod, \
401 SI_SUB_DRIVERS, SI_ORDER_FIRST); \
402 MODULE_DEPEND(g_raid_tr_##name, geom_raid, 0, 0, 0); \
403 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, CTLFLAG_RD, \
404 NULL, label " transformation module"); \
405 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
406 CTLFLAG_RWTUN, &g_raid_tr_##name##_class.trc_enable, 0, \
407 "Enable " label " transformation module taste")
409 const char * g_raid_volume_level2str(int level, int qual);
410 int g_raid_volume_str2level(const char *str, int *level, int *qual);
411 const char * g_raid_volume_state2str(int state);
412 const char * g_raid_subdisk_state2str(int state);
413 const char * g_raid_disk_state2str(int state);
415 struct g_raid_softc * g_raid_create_node(struct g_class *mp,
416 const char *name, struct g_raid_md_object *md);
417 int g_raid_create_node_format(const char *format, struct gctl_req *req,
419 struct g_raid_volume * g_raid_create_volume(struct g_raid_softc *sc,
420 const char *name, int id);
421 struct g_raid_disk * g_raid_create_disk(struct g_raid_softc *sc);
422 const char * g_raid_get_diskname(struct g_raid_disk *disk);
423 void g_raid_get_disk_info(struct g_raid_disk *disk);
425 int g_raid_start_volume(struct g_raid_volume *vol);
427 int g_raid_destroy_node(struct g_raid_softc *sc, int worker);
428 int g_raid_destroy_volume(struct g_raid_volume *vol);
429 int g_raid_destroy_disk(struct g_raid_disk *disk);
431 void g_raid_iodone(struct bio *bp, int error);
432 void g_raid_subdisk_iostart(struct g_raid_subdisk *sd, struct bio *bp);
433 int g_raid_subdisk_kerneldump(struct g_raid_subdisk *sd,
434 void *virtual, vm_offset_t physical, off_t offset, size_t length);
436 struct g_consumer *g_raid_open_consumer(struct g_raid_softc *sc,
438 void g_raid_kill_consumer(struct g_raid_softc *sc, struct g_consumer *cp);
440 void g_raid_report_disk_state(struct g_raid_disk *disk);
441 void g_raid_change_disk_state(struct g_raid_disk *disk, int state);
442 void g_raid_change_subdisk_state(struct g_raid_subdisk *sd, int state);
443 void g_raid_change_volume_state(struct g_raid_volume *vol, int state);
445 void g_raid_write_metadata(struct g_raid_softc *sc, struct g_raid_volume *vol,
446 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
447 void g_raid_fail_disk(struct g_raid_softc *sc,
448 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
450 void g_raid_tr_flush_common(struct g_raid_tr_object *tr, struct bio *bp);
451 int g_raid_tr_kerneldump_common(struct g_raid_tr_object *tr,
452 void *virtual, vm_offset_t physical, off_t offset, size_t length);
454 u_int g_raid_ndisks(struct g_raid_softc *sc, int state);
455 u_int g_raid_nsubdisks(struct g_raid_volume *vol, int state);
456 u_int g_raid_nopens(struct g_raid_softc *sc);
457 struct g_raid_subdisk * g_raid_get_subdisk(struct g_raid_volume *vol,
459 #define G_RAID_DESTROY_SOFT 0
460 #define G_RAID_DESTROY_DELAYED 1
461 #define G_RAID_DESTROY_HARD 2
462 int g_raid_destroy(struct g_raid_softc *sc, int how);
463 int g_raid_event_send(void *arg, int event, int flags);
464 int g_raid_lock_range(struct g_raid_volume *vol, off_t off, off_t len,
465 struct bio *ignore, void *argp);
466 int g_raid_unlock_range(struct g_raid_volume *vol, off_t off, off_t len);
468 g_ctl_req_t g_raid_ctl;
471 #endif /* !_G_RAID_H_ */