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 * $FreeBSD$ 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 62 extern 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_BIO_CFLAG_REGULAR 0x01 88 #define G_RAID3_BIO_CFLAG_SYNC 0x02 89 #define G_RAID3_BIO_CFLAG_PARITY 0x04 90 #define G_RAID3_BIO_CFLAG_NODISK 0x08 91 #define G_RAID3_BIO_CFLAG_REGSYNC 0x10 92 #define G_RAID3_BIO_CFLAG_MASK (G_RAID3_BIO_CFLAG_REGULAR | \ 93 G_RAID3_BIO_CFLAG_SYNC | \ 94 G_RAID3_BIO_CFLAG_PARITY | \ 95 G_RAID3_BIO_CFLAG_NODISK | \ 96 G_RAID3_BIO_CFLAG_REGSYNC) 97 98 #define G_RAID3_BIO_PFLAG_DEGRADED 0x01 99 #define G_RAID3_BIO_PFLAG_NOPARITY 0x02 100 #define G_RAID3_BIO_PFLAG_VERIFY 0x04 101 #define G_RAID3_BIO_PFLAG_MASK (G_RAID3_BIO_PFLAG_DEGRADED | \ 102 G_RAID3_BIO_PFLAG_NOPARITY | \ 103 G_RAID3_BIO_PFLAG_VERIFY) 104 105 /* 106 * Informations needed for synchronization. 107 */ 108 struct g_raid3_disk_sync { 109 struct g_consumer *ds_consumer; /* Consumer connected to our device. */ 110 off_t ds_offset; /* Offset of next request to send. */ 111 off_t ds_offset_done; /* Offset of already synchronized 112 region. */ 113 off_t ds_resync; /* Resynchronize from this offset. */ 114 u_int ds_syncid; /* Disk's synchronization ID. */ 115 u_char *ds_data; 116 }; 117 118 /* 119 * Informations needed for synchronization. 120 */ 121 struct g_raid3_device_sync { 122 struct g_geom *ds_geom; /* Synchronization geom. */ 123 }; 124 125 #define G_RAID3_DISK_STATE_NODISK 0 126 #define G_RAID3_DISK_STATE_NONE 1 127 #define G_RAID3_DISK_STATE_NEW 2 128 #define G_RAID3_DISK_STATE_ACTIVE 3 129 #define G_RAID3_DISK_STATE_STALE 4 130 #define G_RAID3_DISK_STATE_SYNCHRONIZING 5 131 #define G_RAID3_DISK_STATE_DISCONNECTED 6 132 #define G_RAID3_DISK_STATE_DESTROY 7 133 struct g_raid3_disk { 134 u_int d_no; /* Disk number. */ 135 struct g_consumer *d_consumer; /* Consumer. */ 136 struct g_raid3_softc *d_softc; /* Back-pointer to softc. */ 137 int d_state; /* Disk state. */ 138 uint64_t d_flags; /* Additional flags. */ 139 struct g_raid3_disk_sync d_sync; /* Sync information. */ 140 LIST_ENTRY(g_raid3_disk) d_next; 141 }; 142 #define d_name d_consumer->provider->name 143 144 #define G_RAID3_EVENT_DONTWAIT 0x1 145 #define G_RAID3_EVENT_WAIT 0x2 146 #define G_RAID3_EVENT_DEVICE 0x4 147 #define G_RAID3_EVENT_DONE 0x8 148 struct g_raid3_event { 149 struct g_raid3_disk *e_disk; 150 int e_state; 151 int e_flags; 152 int e_error; 153 TAILQ_ENTRY(g_raid3_event) e_next; 154 }; 155 156 #define G_RAID3_DEVICE_FLAG_DESTROY 0x0100000000000000ULL 157 #define G_RAID3_DEVICE_FLAG_WAIT 0x0200000000000000ULL 158 159 #define G_RAID3_DEVICE_STATE_STARTING 0 160 #define G_RAID3_DEVICE_STATE_DEGRADED 1 161 #define G_RAID3_DEVICE_STATE_COMPLETE 2 162 163 #define G_RAID3_BUMP_ON_FIRST_WRITE 1 164 #define G_RAID3_BUMP_IMMEDIATELY 2 165 166 struct g_raid3_softc { 167 u_int sc_state; /* Device state. */ 168 uint64_t sc_mediasize; /* Device size. */ 169 uint32_t sc_sectorsize; /* Sector size. */ 170 uint64_t sc_flags; /* Additional flags. */ 171 172 struct g_geom *sc_geom; 173 struct g_provider *sc_provider; 174 175 uint32_t sc_id; /* Device unique ID. */ 176 177 struct bio_queue_head sc_queue; 178 struct mtx sc_queue_mtx; 179 struct proc *sc_worker; 180 181 struct g_raid3_disk *sc_disks; 182 u_int sc_ndisks; /* Number of disks. */ 183 u_int sc_round_robin; 184 struct g_raid3_disk *sc_syncdisk; 185 186 uma_zone_t sc_zone_64k; 187 uma_zone_t sc_zone_16k; 188 uma_zone_t sc_zone_4k; 189 190 u_int sc_syncid; /* Synchronization ID. */ 191 int sc_bump_syncid; 192 struct g_raid3_device_sync sc_sync; 193 194 TAILQ_HEAD(, g_raid3_event) sc_events; 195 struct mtx sc_events_mtx; 196 197 struct callout sc_callout; 198 }; 199 #define sc_name sc_geom->name 200 201 const char *g_raid3_get_diskname(struct g_raid3_disk *disk); 202 u_int g_raid3_ndisks(struct g_raid3_softc *sc, int state); 203 int g_raid3_destroy(struct g_raid3_softc *sc, boolean_t force); 204 int g_raid3_event_send(void *arg, int state, int flags); 205 struct g_raid3_metadata; 206 void g_raid3_fill_metadata(struct g_raid3_disk *disk, 207 struct g_raid3_metadata *md); 208 int g_raid3_clear_metadata(struct g_raid3_disk *disk); 209 void g_raid3_update_metadata(struct g_raid3_disk *disk); 210 211 g_ctl_req_t g_raid3_config; 212 #endif /* _KERNEL */ 213 214 struct g_raid3_metadata { 215 char md_magic[16]; /* Magic value. */ 216 uint32_t md_version; /* Version number. */ 217 char md_name[16]; /* Device name. */ 218 uint32_t md_id; /* Device unique ID. */ 219 uint16_t md_no; /* Component number. */ 220 uint16_t md_all; /* Number of disks in device. */ 221 uint32_t md_syncid; /* Synchronization ID. */ 222 uint64_t md_mediasize; /* Size of whole device. */ 223 uint32_t md_sectorsize; /* Sector size. */ 224 uint64_t md_sync_offset; /* Synchronized offset. */ 225 uint64_t md_mflags; /* Additional device flags. */ 226 uint64_t md_dflags; /* Additional disk flags. */ 227 char md_provider[16]; /* Hardcoded provider. */ 228 u_char md_hash[16]; /* MD5 hash. */ 229 }; 230 static __inline void 231 raid3_metadata_encode(struct g_raid3_metadata *md, u_char *data) 232 { 233 MD5_CTX ctx; 234 235 bcopy(md->md_magic, data, 16); 236 le32enc(data + 16, md->md_version); 237 bcopy(md->md_name, data + 20, 16); 238 le32enc(data + 36, md->md_id); 239 le16enc(data + 40, md->md_no); 240 le16enc(data + 42, md->md_all); 241 le32enc(data + 44, md->md_syncid); 242 le64enc(data + 48, md->md_mediasize); 243 le32enc(data + 56, md->md_sectorsize); 244 le64enc(data + 60, md->md_sync_offset); 245 le64enc(data + 68, md->md_mflags); 246 le64enc(data + 76, md->md_dflags); 247 bcopy(md->md_provider, data + 84, 16); 248 MD5Init(&ctx); 249 MD5Update(&ctx, data, 100); 250 MD5Final(md->md_hash, &ctx); 251 bcopy(md->md_hash, data + 100, 16); 252 } 253 static __inline int 254 raid3_metadata_decode(const u_char *data, struct g_raid3_metadata *md) 255 { 256 MD5_CTX ctx; 257 258 bcopy(data, md->md_magic, 16); 259 md->md_version = le32dec(data + 16); 260 bcopy(data + 20, md->md_name, 16); 261 md->md_id = le32dec(data + 36); 262 md->md_no = le16dec(data + 40); 263 md->md_all = le16dec(data + 42); 264 md->md_syncid = le32dec(data + 44); 265 md->md_mediasize = le64dec(data + 48); 266 md->md_sectorsize = le32dec(data + 56); 267 md->md_sync_offset = le64dec(data + 60); 268 md->md_mflags = le64dec(data + 68); 269 md->md_dflags = le64dec(data + 76); 270 bcopy(data + 84, md->md_provider, 16); 271 bcopy(data + 100, md->md_hash, 16); 272 MD5Init(&ctx); 273 MD5Update(&ctx, data, 100); 274 MD5Final(md->md_hash, &ctx); 275 if (bcmp(md->md_hash, data + 100, 16) != 0) 276 return (EINVAL); 277 return (0); 278 } 279 280 static __inline void 281 raid3_metadata_dump(const struct g_raid3_metadata *md) 282 { 283 static const char hex[] = "0123456789abcdef"; 284 char hash[16 * 2 + 1]; 285 u_int i; 286 287 printf(" magic: %s\n", md->md_magic); 288 printf(" version: %u\n", (u_int)md->md_version); 289 printf(" name: %s\n", md->md_name); 290 printf(" id: %u\n", (u_int)md->md_id); 291 printf(" no: %u\n", (u_int)md->md_no); 292 printf(" all: %u\n", (u_int)md->md_all); 293 printf(" syncid: %u\n", (u_int)md->md_syncid); 294 printf(" mediasize: %jd\n", (intmax_t)md->md_mediasize); 295 printf("sectorsize: %u\n", (u_int)md->md_sectorsize); 296 printf("syncoffset: %jd\n", (intmax_t)md->md_sync_offset); 297 printf(" mflags:"); 298 if (md->md_mflags == 0) 299 printf(" NONE"); 300 else { 301 if ((md->md_mflags & G_RAID3_DEVICE_FLAG_NOAUTOSYNC) != 0) 302 printf(" NOAUTOSYNC"); 303 if ((md->md_mflags & G_RAID3_DEVICE_FLAG_ROUND_ROBIN) != 0) 304 printf(" ROUND-ROBIN"); 305 if ((md->md_mflags & G_RAID3_DEVICE_FLAG_VERIFY) != 0) 306 printf(" VERIFY"); 307 } 308 printf("\n"); 309 printf(" dflags:"); 310 if (md->md_dflags == 0) 311 printf(" NONE"); 312 else { 313 if ((md->md_dflags & G_RAID3_DISK_FLAG_DIRTY) != 0) 314 printf(" DIRTY"); 315 if ((md->md_dflags & G_RAID3_DISK_FLAG_SYNCHRONIZING) != 0) 316 printf(" SYNCHRONIZING"); 317 if ((md->md_dflags & G_RAID3_DISK_FLAG_FORCE_SYNC) != 0) 318 printf(" FORCE_SYNC"); 319 } 320 printf("\n"); 321 printf("hcprovider: %s\n", md->md_provider); 322 bzero(hash, sizeof(hash)); 323 for (i = 0; i < 16; i++) { 324 hash[i * 2] = hex[md->md_hash[i] >> 4]; 325 hash[i * 2 + 1] = hex[md->md_hash[i] & 0x0f]; 326 } 327 printf(" MD5 hash: %s\n", hash); 328 } 329 #endif /* !_G_RAID3_H_ */ 330