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