2 * SPDX-License-Identifier: BSD-2-Clause
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
32 #include <sys/param.h>
37 #include <sys/sysctl.h>
40 #define G_RAID_CLASS_NAME "RAID"
42 #define G_RAID_MAGIC "GEOM::RAID"
44 #define G_RAID_VERSION 0
46 struct g_raid_md_object;
47 struct g_raid_tr_object;
49 #define G_RAID_DEVICE_FLAG_NOAUTOSYNC 0x0000000000000001ULL
50 #define G_RAID_DEVICE_FLAG_NOFAILSYNC 0x0000000000000002ULL
51 #define G_RAID_DEVICE_FLAG_MASK (G_RAID_DEVICE_FLAG_NOAUTOSYNC | \
52 G_RAID_DEVICE_FLAG_NOFAILSYNC)
55 extern u_int g_raid_aggressive_spare;
56 extern u_int g_raid_debug;
57 extern int g_raid_enable;
58 extern int g_raid_read_err_thresh;
59 extern u_int g_raid_start_timeout;
60 extern struct g_class g_raid_class;
62 #define G_RAID_DEBUG(lvl, ...) \
63 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), NULL, __VA_ARGS__)
64 #define G_RAID_DEBUG1(lvl, sc, fmt, ...) \
65 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), NULL, "%s: " fmt, \
66 (sc)->sc_name, ## __VA_ARGS__)
67 #define G_RAID_LOGREQ(lvl, bp, ...) \
68 _GEOM_DEBUG("GEOM_RAID", g_raid_debug, (lvl), (bp), __VA_ARGS__)
71 * Flags we use to distinguish I/O initiated by the TR layer to maintain
72 * the volume's characteristics, fix subdisks, extra copies of data, etc.
74 * G_RAID_BIO_FLAG_SYNC I/O to update an extra copy of the data
75 * for RAID volumes that maintain extra data
76 * and need to rebuild that data.
77 * G_RAID_BIO_FLAG_REMAP I/O done to try to provoke a subdisk into
78 * doing some desirable action such as bad
79 * block remapping after we detect a bad part
81 * G_RAID_BIO_FLAG_LOCKED I/O holds range lock that should re released.
83 * and the following meta item:
84 * G_RAID_BIO_FLAG_SPECIAL And of the I/O flags that need to make it
85 * through the range locking which would
86 * otherwise defer the I/O until after that
89 #define G_RAID_BIO_FLAG_SYNC 0x01
90 #define G_RAID_BIO_FLAG_REMAP 0x02
91 #define G_RAID_BIO_FLAG_SPECIAL \
92 (G_RAID_BIO_FLAG_SYNC|G_RAID_BIO_FLAG_REMAP)
93 #define G_RAID_BIO_FLAG_LOCKED 0x80
100 LIST_ENTRY(g_raid_lock) l_next;
103 #define G_RAID_EVENT_WAIT 0x01
104 #define G_RAID_EVENT_VOLUME 0x02
105 #define G_RAID_EVENT_SUBDISK 0x04
106 #define G_RAID_EVENT_DISK 0x08
107 #define G_RAID_EVENT_DONE 0x10
108 struct g_raid_event {
113 TAILQ_ENTRY(g_raid_event) e_next;
115 #define G_RAID_DISK_S_NONE 0x00 /* State is unknown. */
116 #define G_RAID_DISK_S_OFFLINE 0x01 /* Missing disk placeholder. */
117 #define G_RAID_DISK_S_DISABLED 0x02 /* Disabled. */
118 #define G_RAID_DISK_S_FAILED 0x03 /* Failed. */
119 #define G_RAID_DISK_S_STALE_FAILED 0x04 /* Old failed. */
120 #define G_RAID_DISK_S_SPARE 0x05 /* Hot-spare. */
121 #define G_RAID_DISK_S_STALE 0x06 /* Old disk, unused now. */
122 #define G_RAID_DISK_S_ACTIVE 0x07 /* Operational. */
124 #define G_RAID_DISK_E_DISCONNECTED 0x01
127 struct g_raid_softc *d_softc; /* Back-pointer to softc. */
128 struct g_consumer *d_consumer; /* GEOM disk consumer. */
129 void *d_md_data; /* Disk's metadata storage. */
130 int d_candelete; /* BIO_DELETE supported. */
131 uint64_t d_flags; /* Additional flags. */
132 u_int d_state; /* Disk state. */
133 u_int d_load; /* Disk average load. */
134 off_t d_last_offset; /* Last head offset. */
135 int d_read_errs; /* Count of the read errors */
136 TAILQ_HEAD(, g_raid_subdisk) d_subdisks; /* List of subdisks. */
137 TAILQ_ENTRY(g_raid_disk) d_next; /* Next disk in the node. */
138 struct g_kerneldump d_kd; /* Kernel dumping method/args. */
141 #define G_RAID_SUBDISK_S_NONE 0x00 /* Absent. */
142 #define G_RAID_SUBDISK_S_FAILED 0x01 /* Failed. */
143 #define G_RAID_SUBDISK_S_NEW 0x02 /* Blank. */
144 #define G_RAID_SUBDISK_S_REBUILD 0x03 /* Blank + rebuild. */
145 #define G_RAID_SUBDISK_S_UNINITIALIZED 0x04 /* Disk of the new volume. */
146 #define G_RAID_SUBDISK_S_STALE 0x05 /* Dirty. */
147 #define G_RAID_SUBDISK_S_RESYNC 0x06 /* Dirty + check/repair. */
148 #define G_RAID_SUBDISK_S_ACTIVE 0x07 /* Usable. */
150 #define G_RAID_SUBDISK_E_NEW 0x01 /* A new subdisk has arrived */
151 #define G_RAID_SUBDISK_E_FAILED 0x02 /* A subdisk failed, but remains in volume */
152 #define G_RAID_SUBDISK_E_DISCONNECTED 0x03 /* A subdisk removed from volume. */
153 #define G_RAID_SUBDISK_E_FIRST_TR_PRIVATE 0x80 /* translation private events */
155 #define G_RAID_SUBDISK_POS(sd) \
156 ((sd)->sd_disk ? ((sd)->sd_disk->d_last_offset - (sd)->sd_offset) : 0)
157 #define G_RAID_SUBDISK_TRACK_SIZE (1 * 1024 * 1024)
158 #define G_RAID_SUBDISK_LOAD(sd) \
159 ((sd)->sd_disk ? ((sd)->sd_disk->d_load) : 0)
160 #define G_RAID_SUBDISK_LOAD_SCALE 256
162 struct g_raid_subdisk {
163 struct g_raid_softc *sd_softc; /* Back-pointer to softc. */
164 struct g_raid_disk *sd_disk; /* Where this subdisk lives. */
165 struct g_raid_volume *sd_volume; /* Volume, sd is a part of. */
166 off_t sd_offset; /* Offset on the disk. */
167 off_t sd_size; /* Size on the disk. */
168 u_int sd_pos; /* Position in volume. */
169 u_int sd_state; /* Subdisk state. */
170 off_t sd_rebuild_pos; /* Rebuild position. */
171 int sd_recovery; /* Count of recovery reqs. */
172 TAILQ_ENTRY(g_raid_subdisk) sd_next; /* Next subdisk on disk. */
175 #define G_RAID_MAX_SUBDISKS 16
176 #define G_RAID_MAX_VOLUMENAME 32
178 #define G_RAID_VOLUME_S_STARTING 0x00
179 #define G_RAID_VOLUME_S_BROKEN 0x01
180 #define G_RAID_VOLUME_S_DEGRADED 0x02
181 #define G_RAID_VOLUME_S_SUBOPTIMAL 0x03
182 #define G_RAID_VOLUME_S_OPTIMAL 0x04
183 #define G_RAID_VOLUME_S_UNSUPPORTED 0x05
184 #define G_RAID_VOLUME_S_STOPPED 0x06
186 #define G_RAID_VOLUME_S_ALIVE(s) \
187 ((s) == G_RAID_VOLUME_S_DEGRADED || \
188 (s) == G_RAID_VOLUME_S_SUBOPTIMAL || \
189 (s) == G_RAID_VOLUME_S_OPTIMAL)
191 #define G_RAID_VOLUME_E_DOWN 0x00
192 #define G_RAID_VOLUME_E_UP 0x01
193 #define G_RAID_VOLUME_E_START 0x10
194 #define G_RAID_VOLUME_E_STARTMD 0x11
196 #define G_RAID_VOLUME_RL_RAID0 0x00
197 #define G_RAID_VOLUME_RL_RAID1 0x01
198 #define G_RAID_VOLUME_RL_RAID3 0x03
199 #define G_RAID_VOLUME_RL_RAID4 0x04
200 #define G_RAID_VOLUME_RL_RAID5 0x05
201 #define G_RAID_VOLUME_RL_RAID6 0x06
202 #define G_RAID_VOLUME_RL_RAIDMDF 0x07
203 #define G_RAID_VOLUME_RL_RAID1E 0x11
204 #define G_RAID_VOLUME_RL_SINGLE 0x0f
205 #define G_RAID_VOLUME_RL_CONCAT 0x1f
206 #define G_RAID_VOLUME_RL_RAID5E 0x15
207 #define G_RAID_VOLUME_RL_RAID5EE 0x25
208 #define G_RAID_VOLUME_RL_RAID5R 0x35
209 #define G_RAID_VOLUME_RL_UNKNOWN 0xff
211 #define G_RAID_VOLUME_RLQ_NONE 0x00
212 #define G_RAID_VOLUME_RLQ_R1SM 0x00
213 #define G_RAID_VOLUME_RLQ_R1MM 0x01
214 #define G_RAID_VOLUME_RLQ_R3P0 0x00
215 #define G_RAID_VOLUME_RLQ_R3PN 0x01
216 #define G_RAID_VOLUME_RLQ_R4P0 0x00
217 #define G_RAID_VOLUME_RLQ_R4PN 0x01
218 #define G_RAID_VOLUME_RLQ_R5RA 0x00
219 #define G_RAID_VOLUME_RLQ_R5RS 0x01
220 #define G_RAID_VOLUME_RLQ_R5LA 0x02
221 #define G_RAID_VOLUME_RLQ_R5LS 0x03
222 #define G_RAID_VOLUME_RLQ_R6RA 0x00
223 #define G_RAID_VOLUME_RLQ_R6RS 0x01
224 #define G_RAID_VOLUME_RLQ_R6LA 0x02
225 #define G_RAID_VOLUME_RLQ_R6LS 0x03
226 #define G_RAID_VOLUME_RLQ_RMDFRA 0x00
227 #define G_RAID_VOLUME_RLQ_RMDFRS 0x01
228 #define G_RAID_VOLUME_RLQ_RMDFLA 0x02
229 #define G_RAID_VOLUME_RLQ_RMDFLS 0x03
230 #define G_RAID_VOLUME_RLQ_R1EA 0x00
231 #define G_RAID_VOLUME_RLQ_R1EO 0x01
232 #define G_RAID_VOLUME_RLQ_R5ERA 0x00
233 #define G_RAID_VOLUME_RLQ_R5ERS 0x01
234 #define G_RAID_VOLUME_RLQ_R5ELA 0x02
235 #define G_RAID_VOLUME_RLQ_R5ELS 0x03
236 #define G_RAID_VOLUME_RLQ_R5EERA 0x00
237 #define G_RAID_VOLUME_RLQ_R5EERS 0x01
238 #define G_RAID_VOLUME_RLQ_R5EELA 0x02
239 #define G_RAID_VOLUME_RLQ_R5EELS 0x03
240 #define G_RAID_VOLUME_RLQ_R5RRA 0x00
241 #define G_RAID_VOLUME_RLQ_R5RRS 0x01
242 #define G_RAID_VOLUME_RLQ_R5RLA 0x02
243 #define G_RAID_VOLUME_RLQ_R5RLS 0x03
244 #define G_RAID_VOLUME_RLQ_UNKNOWN 0xff
246 struct g_raid_volume;
248 struct g_raid_volume {
249 struct g_raid_softc *v_softc; /* Back-pointer to softc. */
250 struct g_provider *v_provider; /* GEOM provider. */
251 struct g_raid_subdisk v_subdisks[G_RAID_MAX_SUBDISKS];
252 /* Subdisks of this volume. */
253 void *v_md_data; /* Volume's metadata storage. */
254 struct g_raid_tr_object *v_tr; /* Transformation object. */
255 char v_name[G_RAID_MAX_VOLUMENAME];
257 u_int v_state; /* Volume state. */
258 u_int v_raid_level; /* Array RAID level. */
259 u_int v_raid_level_qualifier; /* RAID level det. */
260 u_int v_disks_count; /* Number of disks in array. */
261 u_int v_mdf_pdisks; /* Number of parity disks
263 uint16_t v_mdf_polynomial; /* Polynomial for RAIDMDF. */
264 uint8_t v_mdf_method; /* Generation method for RAIDMDF. */
265 u_int v_strip_size; /* Array strip size. */
266 u_int v_rotate_parity; /* Rotate RAID5R parity
267 after numer of stripes. */
268 u_int v_sectorsize; /* Volume sector size. */
269 off_t v_mediasize; /* Volume media size. */
270 struct bio_queue_head v_inflight; /* In-flight write requests. */
271 struct bio_queue_head v_locked; /* Blocked I/O requests. */
272 LIST_HEAD(, g_raid_lock) v_locks; /* List of locked regions. */
273 int v_pending_lock; /* writes to locked region */
274 int v_dirty; /* Volume is DIRTY. */
275 struct timeval v_last_done; /* Time of the last I/O. */
276 time_t v_last_write; /* Time of the last write. */
277 u_int v_writes; /* Number of active writes. */
278 struct root_hold_token *v_rootmount; /* Root mount delay token. */
279 int v_starting; /* Volume is starting */
280 int v_stopping; /* Volume is stopping */
281 int v_provider_open; /* Number of opens. */
282 int v_global_id; /* Global volume ID (rX). */
283 int v_read_only; /* Volume is read-only. */
284 TAILQ_ENTRY(g_raid_volume) v_next; /* List of volumes entry. */
285 LIST_ENTRY(g_raid_volume) v_global_next; /* Global list entry. */
288 #define G_RAID_NODE_E_WAKE 0x00
289 #define G_RAID_NODE_E_START 0x01
291 struct g_raid_softc {
292 struct g_raid_md_object *sc_md; /* Metadata object. */
293 struct g_geom *sc_geom; /* GEOM class instance. */
294 uint64_t sc_flags; /* Additional flags. */
295 TAILQ_HEAD(, g_raid_volume) sc_volumes; /* List of volumes. */
296 TAILQ_HEAD(, g_raid_disk) sc_disks; /* List of disks. */
297 struct sx sc_lock; /* Main node lock. */
298 struct proc *sc_worker; /* Worker process. */
299 struct mtx sc_queue_mtx; /* Worker queues lock. */
300 TAILQ_HEAD(, g_raid_event) sc_events; /* Worker events queue. */
301 struct bio_queue_head sc_queue; /* Worker I/O queue. */
302 int sc_stopping; /* Node is stopping */
304 #define sc_name sc_geom->name
306 SYSCTL_DECL(_kern_geom_raid);
309 * KOBJ parent class of metadata processing modules.
311 struct g_raid_md_class {
315 LIST_ENTRY(g_raid_md_class) mdc_list;
319 * KOBJ instance of metadata processing module.
321 struct g_raid_md_object {
323 struct g_raid_md_class *mdo_class;
324 struct g_raid_softc *mdo_softc; /* Back-pointer to softc. */
327 int g_raid_md_modevent(module_t, int, void *);
329 #define G_RAID_MD_DECLARE(name, label) \
330 static moduledata_t g_raid_md_##name##_mod = { \
331 "g_raid_md_" __XSTRING(name), \
332 g_raid_md_modevent, \
333 &g_raid_md_##name##_class \
335 DECLARE_MODULE(g_raid_md_##name, g_raid_md_##name##_mod, \
336 SI_SUB_DRIVERS, SI_ORDER_SECOND); \
337 MODULE_DEPEND(g_raid_md_##name, geom_raid, 0, 0, 0); \
338 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, \
339 CTLFLAG_RD | CTLFLAG_MPSAFE, \
340 NULL, label " metadata module"); \
341 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
342 CTLFLAG_RWTUN, &g_raid_md_##name##_class.mdc_enable, 0, \
343 "Enable " label " metadata format taste")
346 * KOBJ parent class of data transformation modules.
348 struct g_raid_tr_class {
352 int trc_accept_unmapped;
353 LIST_ENTRY(g_raid_tr_class) trc_list;
357 * KOBJ instance of data transformation module.
359 struct g_raid_tr_object {
361 struct g_raid_tr_class *tro_class;
362 struct g_raid_volume *tro_volume; /* Back-pointer to volume. */
365 int g_raid_tr_modevent(module_t, int, void *);
367 #define G_RAID_TR_DECLARE(name, label) \
368 static moduledata_t g_raid_tr_##name##_mod = { \
369 "g_raid_tr_" __XSTRING(name), \
370 g_raid_tr_modevent, \
371 &g_raid_tr_##name##_class \
373 DECLARE_MODULE(g_raid_tr_##name, g_raid_tr_##name##_mod, \
374 SI_SUB_DRIVERS, SI_ORDER_FIRST); \
375 MODULE_DEPEND(g_raid_tr_##name, geom_raid, 0, 0, 0); \
376 SYSCTL_NODE(_kern_geom_raid, OID_AUTO, name, \
377 CTLFLAG_RD | CTLFLAG_MPSAFE, \
378 NULL, label " transformation module"); \
379 SYSCTL_INT(_kern_geom_raid_##name, OID_AUTO, enable, \
380 CTLFLAG_RWTUN, &g_raid_tr_##name##_class.trc_enable, 0, \
381 "Enable " label " transformation module taste")
383 const char * g_raid_volume_level2str(int level, int qual);
384 int g_raid_volume_str2level(const char *str, int *level, int *qual);
385 const char * g_raid_volume_state2str(int state);
386 const char * g_raid_subdisk_state2str(int state);
387 const char * g_raid_disk_state2str(int state);
389 struct g_raid_softc * g_raid_create_node(struct g_class *mp,
390 const char *name, struct g_raid_md_object *md);
391 int g_raid_create_node_format(const char *format, struct gctl_req *req,
393 struct g_raid_volume * g_raid_create_volume(struct g_raid_softc *sc,
394 const char *name, int id);
395 struct g_raid_disk * g_raid_create_disk(struct g_raid_softc *sc);
396 const char * g_raid_get_diskname(struct g_raid_disk *disk);
397 void g_raid_get_disk_info(struct g_raid_disk *disk);
399 int g_raid_start_volume(struct g_raid_volume *vol);
401 int g_raid_destroy_node(struct g_raid_softc *sc, int worker);
402 int g_raid_destroy_volume(struct g_raid_volume *vol);
403 int g_raid_destroy_disk(struct g_raid_disk *disk);
405 void g_raid_iodone(struct bio *bp, int error);
406 void g_raid_subdisk_iostart(struct g_raid_subdisk *sd, struct bio *bp);
407 int g_raid_subdisk_kerneldump(struct g_raid_subdisk *sd, void *virtual,
408 off_t offset, size_t length);
410 struct g_consumer *g_raid_open_consumer(struct g_raid_softc *sc,
412 void g_raid_kill_consumer(struct g_raid_softc *sc, struct g_consumer *cp);
414 void g_raid_report_disk_state(struct g_raid_disk *disk);
415 void g_raid_change_disk_state(struct g_raid_disk *disk, int state);
416 void g_raid_change_subdisk_state(struct g_raid_subdisk *sd, int state);
417 void g_raid_change_volume_state(struct g_raid_volume *vol, int state);
419 void g_raid_write_metadata(struct g_raid_softc *sc, struct g_raid_volume *vol,
420 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
421 void g_raid_fail_disk(struct g_raid_softc *sc,
422 struct g_raid_subdisk *sd, struct g_raid_disk *disk);
424 void g_raid_tr_flush_common(struct g_raid_tr_object *tr, struct bio *bp);
425 int g_raid_tr_kerneldump_common(struct g_raid_tr_object *tr,
426 void *virtual, vm_offset_t physical, off_t offset, size_t length);
428 u_int g_raid_ndisks(struct g_raid_softc *sc, int state);
429 u_int g_raid_nsubdisks(struct g_raid_volume *vol, int state);
430 u_int g_raid_nopens(struct g_raid_softc *sc);
431 struct g_raid_subdisk * g_raid_get_subdisk(struct g_raid_volume *vol,
433 #define G_RAID_DESTROY_SOFT 0
434 #define G_RAID_DESTROY_DELAYED 1
435 #define G_RAID_DESTROY_HARD 2
436 int g_raid_destroy(struct g_raid_softc *sc, int how);
437 int g_raid_event_send(void *arg, int event, int flags);
438 int g_raid_lock_range(struct g_raid_volume *vol, off_t off, off_t len,
439 struct bio *ignore, void *argp);
440 int g_raid_unlock_range(struct g_raid_volume *vol, off_t off, off_t len);
442 g_ctl_req_t g_raid_ctl;
445 #endif /* !_G_RAID_H_ */