1/*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 2010 Alexander Motin <mav@FreeBSD.org>
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
15 *
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
26 * SUCH DAMAGE.
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD$");
31
32#include <sys/param.h>
33#include <sys/bio.h>
34#include <sys/endian.h>
35#include <sys/kernel.h>
36#include <sys/kobj.h>
37#include <sys/lock.h>
38#include <sys/malloc.h>
39#include <sys/mutex.h>
40#include <sys/systm.h>
41#include <geom/geom.h>
42#include <geom/geom_dbg.h>
43#include "geom/raid/g_raid.h"
44#include "g_raid_tr_if.h"
45
46static MALLOC_DEFINE(M_TR_CONCAT, "tr_concat_data", "GEOM_RAID CONCAT data");
47
48struct g_raid_tr_concat_object {
49	struct g_raid_tr_object	 trso_base;
50	int			 trso_starting;
51	int			 trso_stopped;
52};
53
54static g_raid_tr_taste_t g_raid_tr_taste_concat;
55static g_raid_tr_event_t g_raid_tr_event_concat;
56static g_raid_tr_start_t g_raid_tr_start_concat;
57static g_raid_tr_stop_t g_raid_tr_stop_concat;
58static g_raid_tr_iostart_t g_raid_tr_iostart_concat;
59static g_raid_tr_iodone_t g_raid_tr_iodone_concat;
60static g_raid_tr_kerneldump_t g_raid_tr_kerneldump_concat;
61static g_raid_tr_free_t g_raid_tr_free_concat;
62
63static kobj_method_t g_raid_tr_concat_methods[] = {
64	KOBJMETHOD(g_raid_tr_taste,	g_raid_tr_taste_concat),
65	KOBJMETHOD(g_raid_tr_event,	g_raid_tr_event_concat),
66	KOBJMETHOD(g_raid_tr_start,	g_raid_tr_start_concat),
67	KOBJMETHOD(g_raid_tr_stop,	g_raid_tr_stop_concat),
68	KOBJMETHOD(g_raid_tr_iostart,	g_raid_tr_iostart_concat),
69	KOBJMETHOD(g_raid_tr_iodone,	g_raid_tr_iodone_concat),
70	KOBJMETHOD(g_raid_tr_kerneldump,	g_raid_tr_kerneldump_concat),
71	KOBJMETHOD(g_raid_tr_free,	g_raid_tr_free_concat),
72	{ 0, 0 }
73};
74
75static struct g_raid_tr_class g_raid_tr_concat_class = {
76	"CONCAT",
77	g_raid_tr_concat_methods,
78	sizeof(struct g_raid_tr_concat_object),
79	.trc_enable = 1,
80	.trc_priority = 50,
81	.trc_accept_unmapped = 1
82};
83
84static int
85g_raid_tr_taste_concat(struct g_raid_tr_object *tr, struct g_raid_volume *volume)
86{
87	struct g_raid_tr_concat_object *trs;
88
89	trs = (struct g_raid_tr_concat_object *)tr;
90	if (tr->tro_volume->v_raid_level != G_RAID_VOLUME_RL_SINGLE &&
91	    tr->tro_volume->v_raid_level != G_RAID_VOLUME_RL_CONCAT &&
92	    !(tr->tro_volume->v_disks_count == 1 &&
93	      tr->tro_volume->v_raid_level != G_RAID_VOLUME_RL_UNKNOWN))
94		return (G_RAID_TR_TASTE_FAIL);
95	trs->trso_starting = 1;
96	return (G_RAID_TR_TASTE_SUCCEED);
97}
98
99static int
100g_raid_tr_update_state_concat(struct g_raid_volume *vol)
101{
102	struct g_raid_tr_concat_object *trs;
103	struct g_raid_softc *sc;
104	off_t size;
105	u_int s;
106	int i, n, f;
107
108	sc = vol->v_softc;
109	trs = (struct g_raid_tr_concat_object *)vol->v_tr;
110	if (trs->trso_stopped)
111		s = G_RAID_VOLUME_S_STOPPED;
112	else if (trs->trso_starting)
113		s = G_RAID_VOLUME_S_STARTING;
114	else {
115		n = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_ACTIVE);
116		f = g_raid_nsubdisks(vol, G_RAID_SUBDISK_S_FAILED);
117		if (n + f == vol->v_disks_count) {
118			if (f == 0)
119				s = G_RAID_VOLUME_S_OPTIMAL;
120			else
121				s = G_RAID_VOLUME_S_SUBOPTIMAL;
122		} else
123			s = G_RAID_VOLUME_S_BROKEN;
124	}
125	if (s != vol->v_state) {
126		/*
127		 * Some metadata modules may not know CONCAT volume
128		 * mediasize until all disks connected. Recalculate.
129		 */
130		if (vol->v_raid_level == G_RAID_VOLUME_RL_CONCAT &&
131		    G_RAID_VOLUME_S_ALIVE(s) &&
132		    !G_RAID_VOLUME_S_ALIVE(vol->v_state)) {
133			size = 0;
134			for (i = 0; i < vol->v_disks_count; i++) {
135				if (vol->v_subdisks[i].sd_state !=
136				    G_RAID_SUBDISK_S_NONE)
137					size += vol->v_subdisks[i].sd_size;
138			}
139			vol->v_mediasize = size;
140		}
141
142		g_raid_event_send(vol, G_RAID_VOLUME_S_ALIVE(s) ?
143		    G_RAID_VOLUME_E_UP : G_RAID_VOLUME_E_DOWN,
144		    G_RAID_EVENT_VOLUME);
145		g_raid_change_volume_state(vol, s);
146		if (!trs->trso_starting && !trs->trso_stopped)
147			g_raid_write_metadata(sc, vol, NULL, NULL);
148	}
149	return (0);
150}
151
152static int
153g_raid_tr_event_concat(struct g_raid_tr_object *tr,
154    struct g_raid_subdisk *sd, u_int event)
155{
156	struct g_raid_tr_concat_object *trs;
157	struct g_raid_softc *sc;
158	struct g_raid_volume *vol;
159	int state;
160
161	trs = (struct g_raid_tr_concat_object *)tr;
162	vol = tr->tro_volume;
163	sc = vol->v_softc;
164
165	state = sd->sd_state;
166	if (state != G_RAID_SUBDISK_S_NONE &&
167	    state != G_RAID_SUBDISK_S_FAILED &&
168	    state != G_RAID_SUBDISK_S_ACTIVE) {
169		G_RAID_DEBUG1(1, sc,
170		    "Promote subdisk %s:%d from %s to ACTIVE.",
171		    vol->v_name, sd->sd_pos,
172		    g_raid_subdisk_state2str(sd->sd_state));
173		g_raid_change_subdisk_state(sd, G_RAID_SUBDISK_S_ACTIVE);
174	}
175	if (state != sd->sd_state &&
176	    !trs->trso_starting && !trs->trso_stopped)
177		g_raid_write_metadata(sc, vol, sd, NULL);
178	g_raid_tr_update_state_concat(vol);
179	return (0);
180}
181
182static int
183g_raid_tr_start_concat(struct g_raid_tr_object *tr)
184{
185	struct g_raid_tr_concat_object *trs;
186	struct g_raid_volume *vol;
187
188	trs = (struct g_raid_tr_concat_object *)tr;
189	vol = tr->tro_volume;
190	trs->trso_starting = 0;
191	g_raid_tr_update_state_concat(vol);
192	return (0);
193}
194
195static int
196g_raid_tr_stop_concat(struct g_raid_tr_object *tr)
197{
198	struct g_raid_tr_concat_object *trs;
199	struct g_raid_volume *vol;
200
201	trs = (struct g_raid_tr_concat_object *)tr;
202	vol = tr->tro_volume;
203	trs->trso_starting = 0;
204	trs->trso_stopped = 1;
205	g_raid_tr_update_state_concat(vol);
206	return (0);
207}
208
209static void
210g_raid_tr_iostart_concat(struct g_raid_tr_object *tr, struct bio *bp)
211{
212	struct g_raid_volume *vol;
213	struct g_raid_subdisk *sd;
214	struct bio_queue_head queue;
215	struct bio *cbp;
216	char *addr;
217	off_t offset, length, remain;
218	u_int no;
219
220	vol = tr->tro_volume;
221	if (vol->v_state != G_RAID_VOLUME_S_OPTIMAL &&
222	    vol->v_state != G_RAID_VOLUME_S_SUBOPTIMAL) {
223		g_raid_iodone(bp, EIO);
224		return;
225	}
226	if (bp->bio_cmd == BIO_FLUSH || bp->bio_cmd == BIO_SPEEDUP) {
227		g_raid_tr_flush_common(tr, bp);
228		return;
229	}
230
231	offset = bp->bio_offset;
232	remain = bp->bio_length;
233	if ((bp->bio_flags & BIO_UNMAPPED) != 0)
234		addr = NULL;
235	else
236		addr = bp->bio_data;
237	no = 0;
238	while (no < vol->v_disks_count &&
239	    offset >= vol->v_subdisks[no].sd_size) {
240		offset -= vol->v_subdisks[no].sd_size;
241		no++;
242	}
243	KASSERT(no < vol->v_disks_count,
244	    ("Request starts after volume end (%ju)", bp->bio_offset));
245	bioq_init(&queue);
246	do {
247		sd = &vol->v_subdisks[no];
248		length = MIN(sd->sd_size - offset, remain);
249		cbp = g_clone_bio(bp);
250		if (cbp == NULL)
251			goto failure;
252		cbp->bio_offset = offset;
253		cbp->bio_length = length;
254		if ((bp->bio_flags & BIO_UNMAPPED) != 0 &&
255		    bp->bio_cmd != BIO_DELETE) {
256			cbp->bio_ma_offset += (uintptr_t)addr;
257			cbp->bio_ma += cbp->bio_ma_offset / PAGE_SIZE;
258			cbp->bio_ma_offset %= PAGE_SIZE;
259			cbp->bio_ma_n = round_page(cbp->bio_ma_offset +
260			    cbp->bio_length) / PAGE_SIZE;
261		} else
262			cbp->bio_data = addr;
263		cbp->bio_caller1 = sd;
264		bioq_insert_tail(&queue, cbp);
265		remain -= length;
266		if (bp->bio_cmd != BIO_DELETE)
267			addr += length;
268		offset = 0;
269		no++;
270		KASSERT(no < vol->v_disks_count || remain == 0,
271		    ("Request ends after volume end (%ju, %ju)",
272			bp->bio_offset, bp->bio_length));
273	} while (remain > 0);
274	while ((cbp = bioq_takefirst(&queue)) != NULL) {
275		sd = cbp->bio_caller1;
276		cbp->bio_caller1 = NULL;
277		g_raid_subdisk_iostart(sd, cbp);
278	}
279	return;
280failure:
281	while ((cbp = bioq_takefirst(&queue)) != NULL)
282		g_destroy_bio(cbp);
283	if (bp->bio_error == 0)
284		bp->bio_error = ENOMEM;
285	g_raid_iodone(bp, bp->bio_error);
286}
287
288static int
289g_raid_tr_kerneldump_concat(struct g_raid_tr_object *tr,
290    void *virtual, vm_offset_t physical, off_t boffset, size_t blength)
291{
292	struct g_raid_volume *vol;
293	struct g_raid_subdisk *sd;
294	char *addr;
295	off_t offset, length, remain;
296	int error, no;
297
298	vol = tr->tro_volume;
299	if (vol->v_state != G_RAID_VOLUME_S_OPTIMAL)
300		return (ENXIO);
301
302	offset = boffset;
303	remain = blength;
304	addr = virtual;
305	no = 0;
306	while (no < vol->v_disks_count &&
307	    offset >= vol->v_subdisks[no].sd_size) {
308		offset -= vol->v_subdisks[no].sd_size;
309		no++;
310	}
311	KASSERT(no < vol->v_disks_count,
312	    ("Request starts after volume end (%ju)", boffset));
313	do {
314		sd = &vol->v_subdisks[no];
315		length = MIN(sd->sd_size - offset, remain);
316		error = g_raid_subdisk_kerneldump(&vol->v_subdisks[no],
317		    addr, 0, offset, length);
318		if (error != 0)
319			return (error);
320		remain -= length;
321		addr += length;
322		offset = 0;
323		no++;
324		KASSERT(no < vol->v_disks_count || remain == 0,
325		    ("Request ends after volume end (%ju, %zu)",
326			boffset, blength));
327	} while (remain > 0);
328	return (0);
329}
330
331static void
332g_raid_tr_iodone_concat(struct g_raid_tr_object *tr,
333    struct g_raid_subdisk *sd,struct bio *bp)
334{
335	struct bio *pbp;
336
337	pbp = bp->bio_parent;
338	if (pbp->bio_error == 0)
339		pbp->bio_error = bp->bio_error;
340	g_destroy_bio(bp);
341	pbp->bio_inbed++;
342	if (pbp->bio_children == pbp->bio_inbed) {
343		pbp->bio_completed = pbp->bio_length;
344		g_raid_iodone(pbp, pbp->bio_error);
345	}
346}
347
348static int
349g_raid_tr_free_concat(struct g_raid_tr_object *tr)
350{
351
352	return (0);
353}
354
355G_RAID_TR_DECLARE(concat, "CONCAT");
356