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, ...) \
65 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), NULL, __VA_ARGS__)
66 #define G_RAID_DEBUG1(lvl, sc, fmt, ...) \
67 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), NULL, "%s: " fmt, \
68 (sc)->sc_name, ## __VA_ARGS__)
69 #define G_RAID_LOGREQ(lvl, bp, ...) \
70 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), (bp), __VA_ARGS__)
73 * Flags we use to distinguish I/O initiated by the TR layer to maintain
74 * the volume's characteristics, fix subdisks, extra copies of data, etc.
76 * G_RAID_BIO_FLAG_SYNC I/O to update an extra copy of the data
77 * for RAID volumes that maintain extra data
78 * and need to rebuild that data.
79 * G_RAID_BIO_FLAG_REMAP I/O done to try to provoke a subdisk into
80 * doing some desirable action such as bad
81 * block remapping after we detect a bad part
83 * G_RAID_BIO_FLAG_LOCKED I/O holds range lock that should re released.
85 * and the following meta item:
86 * G_RAID_BIO_FLAG_SPECIAL And of the I/O flags that need to make it
87 * through the range locking which would
88 * otherwise defer the I/O until after that
91 #define G_RAID_BIO_FLAG_SYNC 0x01
92 #define G_RAID_BIO_FLAG_REMAP 0x02
93 #define G_RAID_BIO_FLAG_SPECIAL \
94 (G_RAID_BIO_FLAG_SYNC|G_RAID_BIO_FLAG_REMAP)
95 #define G_RAID_BIO_FLAG_LOCKED 0x80
100 void *l_callback_arg;
102 LIST_ENTRY(g_raid_lock) l_next;
105 #define G_RAID_EVENT_WAIT 0x01
106 #define G_RAID_EVENT_VOLUME 0x02
107 #define G_RAID_EVENT_SUBDISK 0x04
108 #define G_RAID_EVENT_DISK 0x08
109 #define G_RAID_EVENT_DONE 0x10
110 struct g_raid_event {
115 TAILQ_ENTRY(g_raid_event) e_next;
117 #define G_RAID_DISK_S_NONE 0x00 /* State is unknown. */
118 #define G_RAID_DISK_S_OFFLINE 0x01 /* Missing disk placeholder. */
119 #define G_RAID_DISK_S_DISABLED 0x02 /* Disabled. */
120 #define G_RAID_DISK_S_FAILED 0x03 /* Failed. */
121 #define G_RAID_DISK_S_STALE_FAILED 0x04 /* Old failed. */
122 #define G_RAID_DISK_S_SPARE 0x05 /* Hot-spare. */
123 #define G_RAID_DISK_S_STALE 0x06 /* Old disk, unused now. */
124 #define G_RAID_DISK_S_ACTIVE 0x07 /* Operational. */
126 #define G_RAID_DISK_E_DISCONNECTED 0x01
129 struct g_raid_softc *d_softc; /* Back-pointer to softc. */
130 struct g_consumer *d_consumer; /* GEOM disk consumer. */
131 void *d_md_data; /* Disk's metadata storage. */
132 int d_candelete; /* BIO_DELETE supported. */
133 uint64_t d_flags; /* Additional flags. */
134 u_int d_state; /* Disk state. */
135 u_int d_load; /* Disk average load. */
136 off_t d_last_offset; /* Last head offset. */
137 int d_read_errs; /* Count of the read errors */
138 TAILQ_HEAD(, g_raid_subdisk) d_subdisks; /* List of subdisks. */
139 TAILQ_ENTRY(g_raid_disk) d_next; /* Next disk in the node. */
140 struct g_kerneldump d_kd; /* Kernel dumping method/args. */
143 #define G_RAID_SUBDISK_S_NONE 0x00 /* Absent. */
144 #define G_RAID_SUBDISK_S_FAILED 0x01 /* Failed. */
145 #define G_RAID_SUBDISK_S_NEW 0x02 /* Blank. */
146 #define G_RAID_SUBDISK_S_REBUILD 0x03 /* Blank + rebuild. */
147 #define G_RAID_SUBDISK_S_UNINITIALIZED 0x04 /* Disk of the new volume. */
148 #define G_RAID_SUBDISK_S_STALE 0x05 /* Dirty. */
149 #define G_RAID_SUBDISK_S_RESYNC 0x06 /* Dirty + check/repair. */
150 #define G_RAID_SUBDISK_S_ACTIVE 0x07 /* Usable. */
152 #define G_RAID_SUBDISK_E_NEW 0x01 /* A new subdisk has arrived */
153 #define G_RAID_SUBDISK_E_FAILED 0x02 /* A subdisk failed, but remains in volume */
154 #define G_RAID_SUBDISK_E_DISCONNECTED 0x03 /* A subdisk removed from volume. */
155 #define G_RAID_SUBDISK_E_FIRST_TR_PRIVATE 0x80 /* translation private events */
157 #define G_RAID_SUBDISK_POS(sd) \
158 ((sd)->sd_disk ? ((sd)->sd_disk->d_last_offset - (sd)->sd_offset) : 0)
159 #define G_RAID_SUBDISK_TRACK_SIZE (1 * 1024 * 1024)
160 #define G_RAID_SUBDISK_LOAD(sd) \
161 ((sd)->sd_disk ? ((sd)->sd_disk->d_load) : 0)
162 #define G_RAID_SUBDISK_LOAD_SCALE 256
164 struct g_raid_subdisk {
165 struct g_raid_softc *sd_softc; /* Back-pointer to softc. */
166 struct g_raid_disk *sd_disk; /* Where this subdisk lives. */
167 struct g_raid_volume *sd_volume; /* Volume, sd is a part of. */
168 off_t sd_offset; /* Offset on the disk. */
169 off_t sd_size; /* Size on the disk. */
170 u_int sd_pos; /* Position in volume. */
171 u_int sd_state; /* Subdisk state. */
172 off_t sd_rebuild_pos; /* Rebuild position. */
173 int sd_recovery; /* Count of recovery reqs. */
174 TAILQ_ENTRY(g_raid_subdisk) sd_next; /* Next subdisk on disk. */
177 #define G_RAID_MAX_SUBDISKS 16
178 #define G_RAID_MAX_VOLUMENAME 32
180 #define G_RAID_VOLUME_S_STARTING 0x00
181 #define G_RAID_VOLUME_S_BROKEN 0x01
182 #define G_RAID_VOLUME_S_DEGRADED 0x02
183 #define G_RAID_VOLUME_S_SUBOPTIMAL 0x03
184 #define G_RAID_VOLUME_S_OPTIMAL 0x04
185 #define G_RAID_VOLUME_S_UNSUPPORTED 0x05
186 #define G_RAID_VOLUME_S_STOPPED 0x06
188 #define G_RAID_VOLUME_S_ALIVE(s) \
189 ((s) == G_RAID_VOLUME_S_DEGRADED || \
190 (s) == G_RAID_VOLUME_S_SUBOPTIMAL || \
191 (s) == G_RAID_VOLUME_S_OPTIMAL)
193 #define G_RAID_VOLUME_E_DOWN 0x00
194 #define G_RAID_VOLUME_E_UP 0x01
195 #define G_RAID_VOLUME_E_START 0x10
196 #define G_RAID_VOLUME_E_STARTMD 0x11
198 #define G_RAID_VOLUME_RL_RAID0 0x00
199 #define G_RAID_VOLUME_RL_RAID1 0x01
200 #define G_RAID_VOLUME_RL_RAID3 0x03
201 #define G_RAID_VOLUME_RL_RAID4 0x04
202 #define G_RAID_VOLUME_RL_RAID5 0x05
203 #define G_RAID_VOLUME_RL_RAID6 0x06
204 #define G_RAID_VOLUME_RL_RAIDMDF 0x07
205 #define G_RAID_VOLUME_RL_RAID1E 0x11
206 #define G_RAID_VOLUME_RL_SINGLE 0x0f
207 #define G_RAID_VOLUME_RL_CONCAT 0x1f
208 #define G_RAID_VOLUME_RL_RAID5E 0x15
209 #define G_RAID_VOLUME_RL_RAID5EE 0x25
210 #define G_RAID_VOLUME_RL_RAID5R 0x35
211 #define G_RAID_VOLUME_RL_UNKNOWN 0xff
213 #define G_RAID_VOLUME_RLQ_NONE 0x00
214 #define G_RAID_VOLUME_RLQ_R1SM 0x00
215 #define G_RAID_VOLUME_RLQ_R1MM 0x01
216 #define G_RAID_VOLUME_RLQ_R3P0 0x00
217 #define G_RAID_VOLUME_RLQ_R3PN 0x01
218 #define G_RAID_VOLUME_RLQ_R4P0 0x00
219 #define G_RAID_VOLUME_RLQ_R4PN 0x01
220 #define G_RAID_VOLUME_RLQ_R5RA 0x00
221 #define G_RAID_VOLUME_RLQ_R5RS 0x01
222 #define G_RAID_VOLUME_RLQ_R5LA 0x02
223 #define G_RAID_VOLUME_RLQ_R5LS 0x03
224 #define G_RAID_VOLUME_RLQ_R6RA 0x00
225 #define G_RAID_VOLUME_RLQ_R6RS 0x01
226 #define G_RAID_VOLUME_RLQ_R6LA 0x02
227 #define G_RAID_VOLUME_RLQ_R6LS 0x03
228 #define G_RAID_VOLUME_RLQ_RMDFRA 0x00
229 #define G_RAID_VOLUME_RLQ_RMDFRS 0x01
230 #define G_RAID_VOLUME_RLQ_RMDFLA 0x02
231 #define G_RAID_VOLUME_RLQ_RMDFLS 0x03
232 #define G_RAID_VOLUME_RLQ_R1EA 0x00
233 #define G_RAID_VOLUME_RLQ_R1EO 0x01
234 #define G_RAID_VOLUME_RLQ_R5ERA 0x00
235 #define G_RAID_VOLUME_RLQ_R5ERS 0x01
236 #define G_RAID_VOLUME_RLQ_R5ELA 0x02
237 #define G_RAID_VOLUME_RLQ_R5ELS 0x03
238 #define G_RAID_VOLUME_RLQ_R5EERA 0x00
239 #define G_RAID_VOLUME_RLQ_R5EERS 0x01
240 #define G_RAID_VOLUME_RLQ_R5EELA 0x02
241 #define G_RAID_VOLUME_RLQ_R5EELS 0x03
242 #define G_RAID_VOLUME_RLQ_R5RRA 0x00
243 #define G_RAID_VOLUME_RLQ_R5RRS 0x01
244 #define G_RAID_VOLUME_RLQ_R5RLA 0x02
245 #define G_RAID_VOLUME_RLQ_R5RLS 0x03
246 #define G_RAID_VOLUME_RLQ_UNKNOWN 0xff
248 struct g_raid_volume;
250 struct g_raid_volume {
251 struct g_raid_softc *v_softc; /* Back-pointer to softc. */
252 struct g_provider *v_provider; /* GEOM provider. */
253 struct g_raid_subdisk v_subdisks[G_RAID_MAX_SUBDISKS];
254 /* Subdisks of this volume. */
255 void *v_md_data; /* Volume's metadata storage. */
256 struct g_raid_tr_object *v_tr; /* Transformation object. */
257 char v_name[G_RAID_MAX_VOLUMENAME];
259 u_int v_state; /* Volume state. */
260 u_int v_raid_level; /* Array RAID level. */
261 u_int v_raid_level_qualifier; /* RAID level det. */
262 u_int v_disks_count; /* Number of disks in array. */
263 u_int v_mdf_pdisks; /* Number of parity disks
265 uint16_t v_mdf_polynomial; /* Polynomial for RAIDMDF. */
266 uint8_t v_mdf_method; /* Generation method for RAIDMDF. */
267 u_int v_strip_size; /* Array strip size. */
268 u_int v_rotate_parity; /* Rotate RAID5R parity
269 after numer of stripes. */
270 u_int v_sectorsize; /* Volume sector size. */
271 off_t v_mediasize; /* Volume media size. */
272 struct bio_queue_head v_inflight; /* In-flight write requests. */
273 struct bio_queue_head v_locked; /* Blocked I/O requests. */
274 LIST_HEAD(, g_raid_lock) v_locks; /* List of locked regions. */
275 int v_pending_lock; /* writes to locked region */
276 int v_dirty; /* Volume is DIRTY. */
277 struct timeval v_last_done; /* Time of the last I/O. */
278 time_t v_last_write; /* Time of the last write. */
279 u_int v_writes; /* Number of active writes. */
280 struct root_hold_token *v_rootmount; /* Root mount delay token. */
281 int v_starting; /* Volume is starting */
282 int v_stopping; /* Volume is stopping */
283 int v_provider_open; /* Number of opens. */
284 int v_global_id; /* Global volume ID (rX). */
285 int v_read_only; /* Volume is read-only. */
286 TAILQ_ENTRY(g_raid_volume) v_next; /* List of volumes entry. */
287 LIST_ENTRY(g_raid_volume) v_global_next; /* Global list entry. */
290 #define G_RAID_NODE_E_WAKE 0x00
291 #define G_RAID_NODE_E_START 0x01
293 struct g_raid_softc {
294 struct g_raid_md_object *sc_md; /* Metadata object. */
295 struct g_geom *sc_geom; /* GEOM class instance. */
296 uint64_t sc_flags; /* Additional flags. */
297 TAILQ_HEAD(, g_raid_volume) sc_volumes; /* List of volumes. */
298 TAILQ_HEAD(, g_raid_disk) sc_disks; /* List of disks. */
299 struct sx sc_lock; /* Main node lock. */
300 struct proc *sc_worker; /* Worker process. */
301 struct mtx sc_queue_mtx; /* Worker queues lock. */
302 TAILQ_HEAD(, g_raid_event) sc_events; /* Worker events queue. */
303 struct bio_queue_head sc_queue; /* Worker I/O queue. */
304 int sc_stopping; /* Node is stopping */
306 #define sc_name sc_geom->name
308 SYSCTL_DECL(_kern_geom_raid);
311 * KOBJ parent class of metadata processing modules.
313 struct g_raid_md_class {
317 LIST_ENTRY(g_raid_md_class) mdc_list;
321 * KOBJ instance of metadata processing module.
323 struct g_raid_md_object {
325 struct g_raid_md_class *mdo_class;
326 struct g_raid_softc *mdo_softc; /* Back-pointer to softc. */
329 int g_raid_md_modevent(module_t, int, void *);
331 #define G_RAID_MD_DECLARE(name, label) \
332 static moduledata_t g_raid_md_##name##_mod = { \
333 "g_raid_md_" __XSTRING(name), \
334 g_raid_md_modevent, \
335 &g_raid_md_##name##_class \
337 DECLARE_MODULE(g_raid_md_##name, g_raid_md_##name##_mod, \
338 SI_SUB_DRIVERS, SI_ORDER_SECOND); \
339 MODULE_DEPEND(g_raid_md_##name, geom_raid, 0, 0, 0); \
340 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, \
341 CTLFLAG_RD | CTLFLAG_MPSAFE, \
342 NULL, label " metadata module"); \
343 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
344 CTLFLAG_RWTUN, &g_raid_md_##name##_class.mdc_enable, 0, \
345 "Enable " label " metadata format taste")
348 * KOBJ parent class of data transformation modules.
350 struct g_raid_tr_class {
354 int trc_accept_unmapped;
355 LIST_ENTRY(g_raid_tr_class) trc_list;
359 * KOBJ instance of data transformation module.
361 struct g_raid_tr_object {
363 struct g_raid_tr_class *tro_class;
364 struct g_raid_volume *tro_volume; /* Back-pointer to volume. */
367 int g_raid_tr_modevent(module_t, int, void *);
369 #define G_RAID_TR_DECLARE(name, label) \
370 static moduledata_t g_raid_tr_##name##_mod = { \
371 "g_raid_tr_" __XSTRING(name), \
372 g_raid_tr_modevent, \
373 &g_raid_tr_##name##_class \
375 DECLARE_MODULE(g_raid_tr_##name, g_raid_tr_##name##_mod, \
376 SI_SUB_DRIVERS, SI_ORDER_FIRST); \
377 MODULE_DEPEND(g_raid_tr_##name, geom_raid, 0, 0, 0); \
378 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, \
379 CTLFLAG_RD | CTLFLAG_MPSAFE, \
380 NULL, label " transformation module"); \
381 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
382 CTLFLAG_RWTUN, &g_raid_tr_##name##_class.trc_enable, 0, \
383 "Enable " label " transformation module taste")
385 const char * g_raid_volume_level2str(int level, int qual);
386 int g_raid_volume_str2level(const char *str, int *level, int *qual);
387 const char * g_raid_volume_state2str(int state);
388 const char * g_raid_subdisk_state2str(int state);
389 const char * g_raid_disk_state2str(int state);
391 struct g_raid_softc * g_raid_create_node(struct g_class *mp,
392 const char *name, struct g_raid_md_object *md);
393 int g_raid_create_node_format(const char *format, struct gctl_req *req,
395 struct g_raid_volume * g_raid_create_volume(struct g_raid_softc *sc,
396 const char *name, int id);
397 struct g_raid_disk * g_raid_create_disk(struct g_raid_softc *sc);
398 const char * g_raid_get_diskname(struct g_raid_disk *disk);
399 void g_raid_get_disk_info(struct g_raid_disk *disk);
401 int g_raid_start_volume(struct g_raid_volume *vol);
403 int g_raid_destroy_node(struct g_raid_softc *sc, int worker);
404 int g_raid_destroy_volume(struct g_raid_volume *vol);
405 int g_raid_destroy_disk(struct g_raid_disk *disk);
407 void g_raid_iodone(struct bio *bp, int error);
408 void g_raid_subdisk_iostart(struct g_raid_subdisk *sd, struct bio *bp);
409 int g_raid_subdisk_kerneldump(struct g_raid_subdisk *sd, void *virtual,
410 off_t offset, size_t length);
412 struct g_consumer *g_raid_open_consumer(struct g_raid_softc *sc,
414 void g_raid_kill_consumer(struct g_raid_softc *sc, struct g_consumer *cp);
416 void g_raid_report_disk_state(struct g_raid_disk *disk);
417 void g_raid_change_disk_state(struct g_raid_disk *disk, int state);
418 void g_raid_change_subdisk_state(struct g_raid_subdisk *sd, int state);
419 void g_raid_change_volume_state(struct g_raid_volume *vol, int state);
421 void g_raid_write_metadata(struct g_raid_softc *sc, struct g_raid_volume *vol,
422 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
423 void g_raid_fail_disk(struct g_raid_softc *sc,
424 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
426 void g_raid_tr_flush_common(struct g_raid_tr_object *tr, struct bio *bp);
427 int g_raid_tr_kerneldump_common(struct g_raid_tr_object *tr,
428 void *virtual, vm_offset_t physical, off_t offset, size_t length);
430 u_int g_raid_ndisks(struct g_raid_softc *sc, int state);
431 u_int g_raid_nsubdisks(struct g_raid_volume *vol, int state);
432 u_int g_raid_nopens(struct g_raid_softc *sc);
433 struct g_raid_subdisk * g_raid_get_subdisk(struct g_raid_volume *vol,
435 #define G_RAID_DESTROY_SOFT 0
436 #define G_RAID_DESTROY_DELAYED 1
437 #define G_RAID_DESTROY_HARD 2
438 int g_raid_destroy(struct g_raid_softc *sc, int how);
439 int g_raid_event_send(void *arg, int event, int flags);
440 int g_raid_lock_range(struct g_raid_volume *vol, off_t off, off_t len,
441 struct bio *ignore, void *argp);
442 int g_raid_unlock_range(struct g_raid_volume *vol, off_t off, off_t len);
444 g_ctl_req_t g_raid_ctl;
447 #endif /* !_G_RAID_H_ */