g_raid3.h revision 134168 
1/*-
2 * Copyright (c) 2004 Pawel Jakub Dawidek <pjd@FreeBSD.org>
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
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.
13 *
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
24 * SUCH DAMAGE.
25 *
26 * $FreeBSD: head/sys/geom/raid3/g_raid3.h 134168 2004-08-22 16:21:12Z pjd $
27 */
28
29#ifndef	_G_RAID3_H_
30#define	_G_RAID3_H_
31
32#include <sys/endian.h>
33#include <sys/md5.h>
34
35#define	G_RAID3_CLASS_NAME	"RAID3"
36
37#define	G_RAID3_MAGIC		"GEOM::RAID3"
38/*
39 * Version history:
40 * 0 - Initial version number.
41 * 1 - Added 'round-robin reading' algorithm.
42 * 2 - Added 'verify reading' algorithm.
43 */
44#define	G_RAID3_VERSION		2
45
46#define	G_RAID3_DISK_FLAG_DIRTY		0x0000000000000001ULL
47#define	G_RAID3_DISK_FLAG_SYNCHRONIZING	0x0000000000000002ULL
48#define	G_RAID3_DISK_FLAG_FORCE_SYNC	0x0000000000000004ULL
49#define	G_RAID3_DISK_FLAG_HARDCODED	0x0000000000000008ULL
50#define	G_RAID3_DISK_FLAG_MASK		(G_RAID3_DISK_FLAG_DIRTY |	\
51					 G_RAID3_DISK_FLAG_SYNCHRONIZING | \
52					 G_RAID3_DISK_FLAG_FORCE_SYNC)
53
54#define	G_RAID3_DEVICE_FLAG_NOAUTOSYNC	0x0000000000000001ULL
55#define	G_RAID3_DEVICE_FLAG_ROUND_ROBIN	0x0000000000000002ULL
56#define	G_RAID3_DEVICE_FLAG_VERIFY	0x0000000000000004ULL
57#define	G_RAID3_DEVICE_FLAG_MASK	(G_RAID3_DEVICE_FLAG_NOAUTOSYNC | \
58					 G_RAID3_DEVICE_FLAG_ROUND_ROBIN | \
59					 G_RAID3_DEVICE_FLAG_VERIFY)
60
61#ifdef _KERNEL
62extern u_int g_raid3_debug;
63
64#define	G_RAID3_DEBUG(lvl, ...)	do {					\
65	if (g_raid3_debug >= (lvl)) {					\
66		printf("GEOM_RAID3");					\
67		if (g_raid3_debug > 0)					\
68			printf("[%u]", lvl);				\
69		printf(": ");						\
70		printf(__VA_ARGS__);					\
71		printf("\n");						\
72	}								\
73} while (0)
74#define	G_RAID3_LOGREQ(lvl, bp, ...)	do {				\
75	if (g_raid3_debug >= (lvl)) {					\
76		printf("GEOM_RAID3");					\
77		if (g_raid3_debug > 0)					\
78			printf("[%u]", lvl);				\
79		printf(": ");						\
80		printf(__VA_ARGS__);					\
81		printf(" ");						\
82		g_print_bio(bp);					\
83		printf("\n");						\
84	}								\
85} while (0)
86
87#define	G_RAID3_MAX_IO_SIZE	(DFLTPHYS * 2)
88
89#define	G_RAID3_BIO_CFLAG_REGULAR	0x01
90#define	G_RAID3_BIO_CFLAG_SYNC		0x02
91#define	G_RAID3_BIO_CFLAG_PARITY	0x04
92#define	G_RAID3_BIO_CFLAG_NODISK	0x08
93#define	G_RAID3_BIO_CFLAG_REGSYNC	0x10
94#define	G_RAID3_BIO_CFLAG_MASK		(G_RAID3_BIO_CFLAG_REGULAR |	\
95					 G_RAID3_BIO_CFLAG_SYNC |	\
96					 G_RAID3_BIO_CFLAG_PARITY |	\
97					 G_RAID3_BIO_CFLAG_NODISK |	\
98					 G_RAID3_BIO_CFLAG_REGSYNC)
99
100#define	G_RAID3_BIO_PFLAG_DEGRADED	0x01
101#define	G_RAID3_BIO_PFLAG_NOPARITY	0x02
102#define	G_RAID3_BIO_PFLAG_VERIFY	0x04
103#define	G_RAID3_BIO_PFLAG_MASK		(G_RAID3_BIO_PFLAG_DEGRADED |	\
104					 G_RAID3_BIO_PFLAG_NOPARITY |	\
105					 G_RAID3_BIO_PFLAG_VERIFY)
106
107/*
108 * Informations needed for synchronization.
109 */
110struct g_raid3_disk_sync {
111	struct g_consumer *ds_consumer;	/* Consumer connected to our device. */
112	off_t		 ds_offset;	/* Offset of next request to send. */
113	off_t		 ds_offset_done; /* Offset of already synchronized
114					   region. */
115	u_int		 ds_syncid;	/* Disk's synchronization ID. */
116	u_char		*ds_data;
117};
118
119/*
120 * Informations needed for synchronization.
121 */
122struct g_raid3_device_sync {
123	struct g_geom	*ds_geom;	/* Synchronization geom. */
124};
125
126#define	G_RAID3_DISK_STATE_NODISK		0
127#define	G_RAID3_DISK_STATE_NONE			1
128#define	G_RAID3_DISK_STATE_NEW			2
129#define	G_RAID3_DISK_STATE_ACTIVE		3
130#define	G_RAID3_DISK_STATE_STALE		4
131#define	G_RAID3_DISK_STATE_SYNCHRONIZING	5
132#define	G_RAID3_DISK_STATE_DISCONNECTED		6
133#define	G_RAID3_DISK_STATE_DESTROY		7
134struct g_raid3_disk {
135	u_int		 d_no;		/* Disk number. */
136	struct g_consumer *d_consumer;	/* Consumer. */
137	struct g_raid3_softc *d_softc;	/* Back-pointer to softc. */
138	int		 d_state;	/* Disk state. */
139	uint64_t	 d_flags;	/* Additional flags. */
140	struct g_raid3_disk_sync d_sync; /* Sync information. */
141	LIST_ENTRY(g_raid3_disk) d_next;
142};
143#define	d_name	d_consumer->provider->name
144
145#define	G_RAID3_EVENT_DONTWAIT	0x1
146#define	G_RAID3_EVENT_WAIT	0x2
147#define	G_RAID3_EVENT_DEVICE	0x4
148#define	G_RAID3_EVENT_DONE	0x8
149struct g_raid3_event {
150	struct g_raid3_disk	*e_disk;
151	int			 e_state;
152	int			 e_flags;
153	int			 e_error;
154	TAILQ_ENTRY(g_raid3_event) e_next;
155};
156
157#define	G_RAID3_DEVICE_FLAG_DESTROY	0x0100000000000000ULL
158#define	G_RAID3_DEVICE_FLAG_WAIT	0x0200000000000000ULL
159
160#define	G_RAID3_DEVICE_STATE_STARTING		0
161#define	G_RAID3_DEVICE_STATE_DEGRADED		1
162#define	G_RAID3_DEVICE_STATE_COMPLETE		2
163
164#define	G_RAID3_BUMP_ON_FIRST_WRITE		1
165#define	G_RAID3_BUMP_IMMEDIATELY		2
166
167struct g_raid3_softc {
168	u_int		sc_state;	/* Device state. */
169	uint64_t	sc_mediasize;	/* Device size. */
170	uint32_t	sc_sectorsize;	/* Sector size. */
171	uint64_t	sc_flags;	/* Additional flags. */
172
173	struct g_geom	*sc_geom;
174	struct g_provider *sc_provider;
175
176	uint32_t	sc_id;		/* Device unique ID. */
177
178	struct bio_queue_head sc_queue;
179	struct mtx	 sc_queue_mtx;
180	struct proc	*sc_worker;
181
182	struct g_raid3_disk *sc_disks;
183	u_int		sc_ndisks;	/* Number of disks. */
184	u_int		sc_round_robin;
185	struct g_raid3_disk *sc_syncdisk;
186
187	uma_zone_t	sc_zone_64k;
188	uma_zone_t	sc_zone_16k;
189	uma_zone_t	sc_zone_4k;
190
191	u_int		sc_syncid;	/* Synchronization ID. */
192	int		sc_bump_syncid;
193	struct g_raid3_device_sync sc_sync;
194
195	TAILQ_HEAD(, g_raid3_event) sc_events;
196	struct mtx	sc_events_mtx;
197
198	struct callout	sc_callout;
199};
200#define	sc_name	sc_geom->name
201
202const char *g_raid3_get_diskname(struct g_raid3_disk *disk);
203u_int g_raid3_ndisks(struct g_raid3_softc *sc, int state);
204int g_raid3_destroy(struct g_raid3_softc *sc, boolean_t force);
205int g_raid3_event_send(void *arg, int state, int flags);
206struct g_raid3_metadata;
207void g_raid3_fill_metadata(struct g_raid3_disk *disk,
208    struct g_raid3_metadata *md);
209int g_raid3_clear_metadata(struct g_raid3_disk *disk);
210void g_raid3_update_metadata(struct g_raid3_disk *disk);
211
212g_ctl_req_t g_raid3_config;
213#endif	/* _KERNEL */
214
215struct g_raid3_metadata {
216	char		md_magic[16];	/* Magic value. */
217	uint32_t	md_version;	/* Version number. */
218	char		md_name[16];	/* Device name. */
219	uint32_t	md_id;		/* Device unique ID. */
220	uint16_t	md_no;		/* Component number. */
221	uint16_t	md_all;		/* Number of disks in device. */
222	uint32_t	md_syncid;	/* Synchronization ID. */
223	uint64_t	md_mediasize;	/* Size of whole device. */
224	uint32_t	md_sectorsize;	/* Sector size. */
225	uint64_t	md_sync_offset;	/* Synchronized offset. */
226	uint64_t	md_mflags;	/* Additional device flags. */
227	uint64_t	md_dflags;	/* Additional disk flags. */
228	char		md_provider[16]; /* Hardcoded provider. */
229	u_char		md_hash[16];	/* MD5 hash. */
230};
231static __inline void
232raid3_metadata_encode(struct g_raid3_metadata *md, u_char *data)
233{
234	MD5_CTX ctx;
235
236	bcopy(md->md_magic, data, 16);
237	le32enc(data + 16, md->md_version);
238	bcopy(md->md_name, data + 20, 16);
239	le32enc(data + 36, md->md_id);
240	le16enc(data + 40, md->md_no);
241	le16enc(data + 42, md->md_all);
242	le32enc(data + 44, md->md_syncid);
243	le64enc(data + 48, md->md_mediasize);
244	le32enc(data + 56, md->md_sectorsize);
245	le64enc(data + 60, md->md_sync_offset);
246	le64enc(data + 68, md->md_mflags);
247	le64enc(data + 76, md->md_dflags);
248	bcopy(md->md_provider, data + 84, 16);
249	MD5Init(&ctx);
250	MD5Update(&ctx, data, 100);
251	MD5Final(md->md_hash, &ctx);
252	bcopy(md->md_hash, data + 100, 16);
253}
254static __inline int
255raid3_metadata_decode(const u_char *data, struct g_raid3_metadata *md)
256{
257	MD5_CTX ctx;
258
259	bcopy(data, md->md_magic, 16);
260	md->md_version = le32dec(data + 16);
261	bcopy(data + 20, md->md_name, 16);
262	md->md_id = le32dec(data + 36);
263	md->md_no = le16dec(data + 40);
264	md->md_all = le16dec(data + 42);
265	md->md_syncid = le32dec(data + 44);
266	md->md_mediasize = le64dec(data + 48);
267	md->md_sectorsize = le32dec(data + 56);
268	md->md_sync_offset = le64dec(data + 60);
269	md->md_mflags = le64dec(data + 68);
270	md->md_dflags = le64dec(data + 76);
271	bcopy(data + 84, md->md_provider, 16);
272	bcopy(data + 100, md->md_hash, 16);
273	MD5Init(&ctx);
274	MD5Update(&ctx, data, 100);
275	MD5Final(md->md_hash, &ctx);
276	if (bcmp(md->md_hash, data + 100, 16) != 0)
277		return (EINVAL);
278	return (0);
279}
280
281static __inline void
282raid3_metadata_dump(const struct g_raid3_metadata *md)
283{
284	static const char hex[] = "0123456789abcdef";
285	char hash[16 * 2 + 1];
286	u_int i;
287
288	printf("     magic: %s\n", md->md_magic);
289	printf("   version: %u\n", (u_int)md->md_version);
290	printf("      name: %s\n", md->md_name);
291	printf("        id: %u\n", (u_int)md->md_id);
292	printf("        no: %u\n", (u_int)md->md_no);
293	printf("       all: %u\n", (u_int)md->md_all);
294	printf("    syncid: %u\n", (u_int)md->md_syncid);
295	printf(" mediasize: %jd\n", (intmax_t)md->md_mediasize);
296	printf("sectorsize: %u\n", (u_int)md->md_sectorsize);
297	printf("syncoffset: %jd\n", (intmax_t)md->md_sync_offset);
298	printf("    mflags:");
299	if (md->md_mflags == 0)
300		printf(" NONE");
301	else {
302		if ((md->md_mflags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) != 0)
303			printf(" NOAUTOSYNC");
304		if ((md->md_mflags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0)
305			printf(" ROUND-ROBIN");
306		if ((md->md_mflags & G_RAID3_DEVICE_FLAG_VERIFY) != 0)
307			printf(" VERIFY");
308	}
309	printf("\n");
310	printf("    dflags:");
311	if (md->md_dflags == 0)
312		printf(" NONE");
313	else {
314		if ((md->md_dflags & G_RAID3_DISK_FLAG_DIRTY) != 0)
315			printf(" DIRTY");
316		if ((md->md_dflags & G_RAID3_DISK_FLAG_SYNCHRONIZING) != 0)
317			printf(" SYNCHRONIZING");
318		if ((md->md_dflags & G_RAID3_DISK_FLAG_FORCE_SYNC) != 0)
319			printf(" FORCE_SYNC");
320	}
321	printf("\n");
322	printf("hcprovider: %s\n", md->md_provider);
323	bzero(hash, sizeof(hash));
324	for (i = 0; i < 16; i++) {
325		hash[i * 2] = hex[md->md_hash[i] >> 4];
326		hash[i * 2 + 1] = hex[md->md_hash[i] & 0x0f];
327	}
328	printf("  MD5 hash: %s\n", hash);
329}
330#endif	/* !_G_RAID3_H_ */
331