2 * Copyright (c) 2012 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
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
30 #include <sys/param.h>
32 #include <sys/endian.h>
33 #include <sys/kernel.h>
35 #include <sys/limits.h>
37 #include <sys/malloc.h>
38 #include <sys/mutex.h>
39 #include <sys/sysctl.h>
40 #include <sys/systm.h>
41 #include <geom/geom.h>
42 #include "geom/raid/g_raid.h"
43 #include "g_raid_tr_if.h"
45 static MALLOC_DEFINE(M_TR_RAID5, "tr_raid5_data", "GEOM_RAID RAID5 data");
47 #define TR_RAID5_NONE 0
48 #define TR_RAID5_REBUILD 1
49 #define TR_RAID5_RESYNC 2
51 #define TR_RAID5_F_DOING_SOME 0x1
52 #define TR_RAID5_F_LOCKED 0x2
53 #define TR_RAID5_F_ABORT 0x4
55 struct g_raid_tr_raid5_object {
56 struct g_raid_tr_object trso_base;
60 int trso_recover_slabs; /* slabs before rest */
64 struct g_raid_subdisk *trso_failed_sd; /* like per volume */
65 void *trso_buffer; /* Buffer space */
69 static g_raid_tr_taste_t g_raid_tr_taste_raid5;
70 static g_raid_tr_event_t g_raid_tr_event_raid5;
71 static g_raid_tr_start_t g_raid_tr_start_raid5;
72 static g_raid_tr_stop_t g_raid_tr_stop_raid5;
73 static g_raid_tr_iostart_t g_raid_tr_iostart_raid5;
74 static g_raid_tr_iodone_t g_raid_tr_iodone_raid5;
75 static g_raid_tr_kerneldump_t g_raid_tr_kerneldump_raid5;
76 static g_raid_tr_locked_t g_raid_tr_locked_raid5;
77 static g_raid_tr_free_t g_raid_tr_free_raid5;
79 static kobj_method_t g_raid_tr_raid5_methods[] = {
80 KOBJMETHOD(g_raid_tr_taste, g_raid_tr_taste_raid5),
81 KOBJMETHOD(g_raid_tr_event, g_raid_tr_event_raid5),
82 KOBJMETHOD(g_raid_tr_start, g_raid_tr_start_raid5),
83 KOBJMETHOD(g_raid_tr_stop, g_raid_tr_stop_raid5),
84 KOBJMETHOD(g_raid_tr_iostart, g_raid_tr_iostart_raid5),
85 KOBJMETHOD(g_raid_tr_iodone, g_raid_tr_iodone_raid5),
86 KOBJMETHOD(g_raid_tr_kerneldump, g_raid_tr_kerneldump_raid5),
87 KOBJMETHOD(g_raid_tr_locked, g_raid_tr_locked_raid5),
88 KOBJMETHOD(g_raid_tr_free, g_raid_tr_free_raid5),
92 static struct g_raid_tr_class g_raid_tr_raid5_class = {
94 g_raid_tr_raid5_methods,
95 sizeof(struct g_raid_tr_raid5_object),
101 g_raid_tr_taste_raid5(struct g_raid_tr_object *tr, struct g_raid_volume *vol)
103 struct g_raid_tr_raid5_object *trs;
106 trs = (struct g_raid_tr_raid5_object *)tr;
107 qual = tr->tro_volume->v_raid_level_qualifier;
108 if (tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID4 &&
109 (qual == G_RAID_VOLUME_RLQ_R4P0 ||
110 qual == G_RAID_VOLUME_RLQ_R4PN)) {
112 } else if ((tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID5 ||
113 tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID5E ||
114 tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID5EE ||
115 tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID5R ||
116 tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAID6 ||
117 tr->tro_volume->v_raid_level == G_RAID_VOLUME_RL_RAIDMDF) &&
118 (qual == G_RAID_VOLUME_RLQ_R5RA ||
119 qual == G_RAID_VOLUME_RLQ_R5RS ||
120 qual == G_RAID_VOLUME_RLQ_R5LA ||
121 qual == G_RAID_VOLUME_RLQ_R5LS)) {
122 /* RAID5/5E/5EE/5R/6/MDF */
124 return (G_RAID_TR_TASTE_FAIL);
125 trs->trso_starting = 1;
126 return (G_RAID_TR_TASTE_SUCCEED);
130 g_raid_tr_update_state_raid5(struct g_raid_volume *vol,
131 struct g_raid_subdisk *sd)
133 struct g_raid_tr_raid5_object *trs;
134 struct g_raid_softc *sc;
139 trs = (struct g_raid_tr_raid5_object *)vol->v_tr;
140 if (trs->trso_stopping &&
141 (trs->trso_flags & TR_RAID5_F_DOING_SOME) == 0)
142 s = G_RAID_VOLUME_S_STOPPED;
143 else if (trs->trso_starting)
144 s = G_RAID_VOLUME_S_STARTING;
146 na = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_ACTIVE);
147 ns = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_STALE) +
148 g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_RESYNC);
149 nu = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_UNINITIALIZED);
150 if (na == vol->v_disks_count)
151 s = G_RAID_VOLUME_S_OPTIMAL;
152 else if (na + ns == vol->v_disks_count ||
153 na + ns + nu == vol->v_disks_count /* XXX: Temporary. */)
154 s = G_RAID_VOLUME_S_SUBOPTIMAL;
155 else if (na == vol->v_disks_count - 1 ||
156 na + ns + nu == vol->v_disks_count)
157 s = G_RAID_VOLUME_S_DEGRADED;
159 s = G_RAID_VOLUME_S_BROKEN;
161 if (s != vol->v_state) {
162 g_raid_event_send(vol, G_RAID_VOLUME_S_ALIVE(s) ?
163 G_RAID_VOLUME_E_UP : G_RAID_VOLUME_E_DOWN,
164 G_RAID_EVENT_VOLUME);
165 g_raid_change_volume_state(vol, s);
166 if (!trs->trso_starting && !trs->trso_stopping)
167 g_raid_write_metadata(sc, vol, NULL, NULL);
173 g_raid_tr_event_raid5(struct g_raid_tr_object *tr,
174 struct g_raid_subdisk *sd, u_int event)
177 g_raid_tr_update_state_raid5(tr->tro_volume, sd);
182 g_raid_tr_start_raid5(struct g_raid_tr_object *tr)
184 struct g_raid_tr_raid5_object *trs;
185 struct g_raid_volume *vol;
187 trs = (struct g_raid_tr_raid5_object *)tr;
188 trs->trso_starting = 0;
189 vol = tr->tro_volume;
190 vol->v_read_only = 1;
191 g_raid_tr_update_state_raid5(vol, NULL);
196 g_raid_tr_stop_raid5(struct g_raid_tr_object *tr)
198 struct g_raid_tr_raid5_object *trs;
199 struct g_raid_volume *vol;
201 trs = (struct g_raid_tr_raid5_object *)tr;
202 vol = tr->tro_volume;
203 trs->trso_starting = 0;
204 trs->trso_stopping = 1;
205 g_raid_tr_update_state_raid5(vol, NULL);
210 g_raid_tr_iostart_raid5_read(struct g_raid_tr_object *tr, struct bio *bp)
212 struct g_raid_volume *vol;
213 struct g_raid_subdisk *sd;
214 struct bio_queue_head queue;
217 off_t offset, start, length, nstripe, remain;
218 int no, pno, ddisks, pdisks, protate, pleft;
219 u_int strip_size, lvl, qual;
221 vol = tr->tro_volume;
223 strip_size = vol->v_strip_size;
224 lvl = tr->tro_volume->v_raid_level;
225 qual = tr->tro_volume->v_raid_level_qualifier;
226 protate = tr->tro_volume->v_rotate_parity;
229 nstripe = bp->bio_offset / strip_size;
230 /* Start position in stripe. */
231 start = bp->bio_offset % strip_size;
232 /* Number of data and parity disks. */
233 if (lvl == G_RAID_VOLUME_RL_RAIDMDF)
234 pdisks = tr->tro_volume->v_mdf_pdisks;
235 else if (lvl == G_RAID_VOLUME_RL_RAID5EE ||
236 lvl == G_RAID_VOLUME_RL_RAID6)
240 ddisks = vol->v_disks_count - pdisks;
241 /* Parity disk number. */
242 if (lvl == G_RAID_VOLUME_RL_RAID4) {
243 if (qual == 0) /* P0 */
249 pno = (nstripe / (ddisks * protate)) % vol->v_disks_count;
250 pleft = protate - (nstripe / ddisks) % protate;
251 if (qual >= 2) { /* PN/Left */
254 pno += vol->v_disks_count;
257 /* Data disk number. */
258 no = nstripe % ddisks;
259 if (lvl == G_RAID_VOLUME_RL_RAID4) {
262 } else if (qual & 1) { /* Continuation/Symmetric */
263 no = (pno + pdisks + no) % vol->v_disks_count;
264 } else if (no >= pno) /* Restart/Asymmetric */
267 no += imax(0, pno + pdisks - vol->v_disks_count);
268 /* Stripe start position in disk. */
269 offset = (nstripe / ddisks) * strip_size;
270 /* Length of data to operate. */
271 remain = bp->bio_length;
275 length = MIN(strip_size - start, remain);
276 cbp = g_clone_bio(bp);
279 cbp->bio_offset = offset + start;
280 cbp->bio_data = addr;
281 cbp->bio_length = length;
282 cbp->bio_caller1 = &vol->v_subdisks[no];
283 bioq_insert_tail(&queue, cbp);
285 if (lvl == G_RAID_VOLUME_RL_RAID4) {
286 no %= vol->v_disks_count;
288 no = (no + pdisks) % vol->v_disks_count;
289 } else if (qual & 1) { /* Continuation/Symmetric */
290 no %= vol->v_disks_count;
292 if ((--pleft) <= 0) {
294 if (qual < 2) /* P0/Right */
297 pno += vol->v_disks_count - 1;
298 pno %= vol->v_disks_count;
300 no = (pno + pdisks) % vol->v_disks_count;
301 offset += strip_size;
303 } else { /* Restart/Asymmetric */
306 if (no >= vol->v_disks_count) {
307 no -= vol->v_disks_count;
308 if ((--pleft) <= 0) {
310 if (qual < 2) /* P0/Right */
313 pno += vol->v_disks_count - 1;
314 pno %= vol->v_disks_count;
319 no += imax(0, pno + pdisks - vol->v_disks_count);
320 offset += strip_size;
326 } while (remain > 0);
327 for (cbp = bioq_first(&queue); cbp != NULL;
328 cbp = bioq_first(&queue)) {
329 bioq_remove(&queue, cbp);
330 sd = cbp->bio_caller1;
331 cbp->bio_caller1 = NULL;
332 g_raid_subdisk_iostart(sd, cbp);
336 for (cbp = bioq_first(&queue); cbp != NULL;
337 cbp = bioq_first(&queue)) {
338 bioq_remove(&queue, cbp);
341 if (bp->bio_error == 0)
342 bp->bio_error = ENOMEM;
343 g_raid_iodone(bp, bp->bio_error);
347 g_raid_tr_iostart_raid5(struct g_raid_tr_object *tr, struct bio *bp)
349 struct g_raid_volume *vol;
350 struct g_raid_tr_raid5_object *trs;
352 vol = tr->tro_volume;
353 trs = (struct g_raid_tr_raid5_object *)tr;
354 if (vol->v_state < G_RAID_VOLUME_S_SUBOPTIMAL) {
355 g_raid_iodone(bp, EIO);
358 switch (bp->bio_cmd) {
360 g_raid_tr_iostart_raid5_read(tr, bp);
365 g_raid_iodone(bp, ENODEV);
368 KASSERT(1 == 0, ("Invalid command here: %u (volume=%s)",
369 bp->bio_cmd, vol->v_name));
375 g_raid_tr_iodone_raid5(struct g_raid_tr_object *tr,
376 struct g_raid_subdisk *sd, struct bio *bp)
381 pbp = bp->bio_parent;
383 error = bp->bio_error;
385 if (pbp->bio_children == pbp->bio_inbed) {
386 pbp->bio_completed = pbp->bio_length;
387 g_raid_iodone(pbp, error);
392 g_raid_tr_kerneldump_raid5(struct g_raid_tr_object *tr,
393 void *virtual, vm_offset_t physical, off_t offset, size_t length)
400 g_raid_tr_locked_raid5(struct g_raid_tr_object *tr, void *argp)
403 struct g_raid_subdisk *sd;
405 bp = (struct bio *)argp;
406 sd = (struct g_raid_subdisk *)bp->bio_caller1;
407 g_raid_subdisk_iostart(sd, bp);
413 g_raid_tr_free_raid5(struct g_raid_tr_object *tr)
415 struct g_raid_tr_raid5_object *trs;
417 trs = (struct g_raid_tr_raid5_object *)tr;
419 if (trs->trso_buffer != NULL) {
420 free(trs->trso_buffer, M_TR_RAID5);
421 trs->trso_buffer = NULL;
426 G_RAID_TR_DECLARE(raid5, "RAID5");