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_MIRROR_H_ 30 #define _G_MIRROR_H_ 31 32 #include <sys/endian.h> 33 #include <sys/md5.h> 34 35 #define G_MIRROR_CLASS_NAME "MIRROR" 36 37 #define G_MIRROR_MAGIC "GEOM::MIRROR" 38 #define G_MIRROR_VERSION 0 39 40 #define G_MIRROR_BALANCE_NONE 0 41 #define G_MIRROR_BALANCE_ROUND_ROBIN 1 42 #define G_MIRROR_BALANCE_LOAD 2 43 #define G_MIRROR_BALANCE_SPLIT 3 44 #define G_MIRROR_BALANCE_MIN G_MIRROR_BALANCE_NONE 45 #define G_MIRROR_BALANCE_MAX G_MIRROR_BALANCE_SPLIT 46 47 #define G_MIRROR_DISK_FLAG_DIRTY 0x0000000000000001ULL 48 #define G_MIRROR_DISK_FLAG_SYNCHRONIZING 0x0000000000000002ULL 49 #define G_MIRROR_DISK_FLAG_FORCE_SYNC 0x0000000000000004ULL 50 #define G_MIRROR_DISK_FLAG_INACTIVE 0x0000000000000008ULL 51 #define G_MIRROR_DISK_FLAG_MASK (G_MIRROR_DISK_FLAG_DIRTY | \ 52 G_MIRROR_DISK_FLAG_SYNCHRONIZING | \ 53 G_MIRROR_DISK_FLAG_FORCE_SYNC | \ 54 G_MIRROR_DISK_FLAG_INACTIVE) 55 56 #define G_MIRROR_DEVICE_FLAG_NOAUTOSYNC 0x0000000000000001ULL 57 #define G_MIRROR_DEVICE_FLAG_MASK (G_MIRROR_DEVICE_FLAG_NOAUTOSYNC) 58 59 #ifdef _KERNEL 60 extern u_int g_mirror_debug; 61 62 #define G_MIRROR_DEBUG(lvl, ...) do { \ 63 if (g_mirror_debug >= (lvl)) { \ 64 printf("GEOM_MIRROR"); \ 65 if (g_mirror_debug > 0) \ 66 printf("[%u]", lvl); \ 67 printf(": "); \ 68 printf(__VA_ARGS__); \ 69 printf("\n"); \ 70 } \ 71 } while (0) 72 #define G_MIRROR_LOGREQ(lvl, bp, ...) do { \ 73 if (g_mirror_debug >= (lvl)) { \ 74 printf("GEOM_MIRROR"); \ 75 if (g_mirror_debug > 0) \ 76 printf("[%u]", lvl); \ 77 printf(": "); \ 78 printf(__VA_ARGS__); \ 79 printf(" "); \ 80 g_print_bio(bp); \ 81 printf("\n"); \ 82 } \ 83 } while (0) 84 85 /* 86 * Informations needed for synchronization. 87 */ 88 struct g_mirror_disk_sync { 89 struct g_consumer *ds_consumer; /* Consumer connected to our mirror. */ 90 off_t ds_offset; /* Offset of next request to send. */ 91 off_t ds_offset_done; /* Offset of already synchronized 92 region. */ 93 u_int ds_syncid; /* Disk's synchronization ID. */ 94 }; 95 96 /* 97 * Informations needed for synchronization. 98 */ 99 struct g_mirror_device_sync { 100 struct g_geom *ds_geom; /* Synchronization geom. */ 101 size_t ds_block; /* Synchronization request size. */ 102 u_int ds_ndisks; /* Number of disks in SYNCHRONIZING 103 state. */ 104 uma_zone_t ds_zone; /* UMA zone for synchronization 105 blocks. */ 106 }; 107 108 #define G_MIRROR_DISK_STATE_NONE 0 109 #define G_MIRROR_DISK_STATE_NEW 1 110 #define G_MIRROR_DISK_STATE_ACTIVE 2 111 #define G_MIRROR_DISK_STATE_STALE 3 112 #define G_MIRROR_DISK_STATE_SYNCHRONIZING 4 113 #define G_MIRROR_DISK_STATE_DISCONNECTED 5 114 #define G_MIRROR_DISK_STATE_DESTROY 6 115 struct g_mirror_disk { 116 uint32_t d_id; /* Disk ID. */ 117 struct g_consumer *d_consumer; /* Consumer. */ 118 struct g_mirror_softc *d_softc; /* Back-pointer to softc. */ 119 int d_state; /* Disk state. */ 120 u_int d_priority; /* Disk priority. */ 121 struct bintime d_delay; /* Disk delay. */ 122 struct bintime d_last_used; /* When disk was last used. */ 123 uint64_t d_flags; /* Additional flags. */ 124 struct g_mirror_disk_sync d_sync;/* Sync information. */ 125 LIST_ENTRY(g_mirror_disk) d_next; 126 }; 127 #define d_name d_consumer->provider->name 128 129 #define G_MIRROR_EVENT_DONTWAIT 0x1 130 #define G_MIRROR_EVENT_WAIT 0x2 131 #define G_MIRROR_EVENT_DEVICE 0x4 132 #define G_MIRROR_EVENT_DONE 0x8 133 struct g_mirror_event { 134 struct g_mirror_disk *e_disk; 135 int e_state; 136 int e_flags; 137 int e_error; 138 TAILQ_ENTRY(g_mirror_event) e_next; 139 }; 140 141 #define G_MIRROR_DEVICE_FLAG_DESTROY 0x0100000000000000ULL 142 #define G_MIRROR_DEVICE_FLAG_WAIT 0x0200000000000000ULL 143 144 #define G_MIRROR_DEVICE_STATE_STARTING 0 145 #define G_MIRROR_DEVICE_STATE_RUNNING 1 146 147 #define G_MIRROR_BUMP_ON_FIRST_WRITE 1 148 #define G_MIRROR_BUMP_IMMEDIATELY 2 149 struct g_mirror_softc { 150 u_int sc_state; /* Device state. */ 151 uint32_t sc_slice; /* Slice size. */ 152 uint8_t sc_balance; /* Balance algorithm. */ 153 uint64_t sc_mediasize; /* Device size. */ 154 uint32_t sc_sectorsize; /* Sector size. */ 155 uint64_t sc_flags; /* Additional flags. */ 156 157 struct g_geom *sc_geom; 158 struct g_provider *sc_provider; 159 160 uint32_t sc_id; /* Mirror unique ID. */ 161 162 struct bio_queue_head sc_queue; 163 struct mtx sc_queue_mtx; 164 struct proc *sc_worker; 165 166 LIST_HEAD(, g_mirror_disk) sc_disks; 167 u_int sc_ndisks; /* Number of disks. */ 168 struct g_mirror_disk *sc_hint; 169 170 u_int sc_syncid; /* Synchronization ID. */ 171 int sc_bump_syncid; 172 struct g_mirror_device_sync sc_sync; 173 174 TAILQ_HEAD(, g_mirror_event) sc_events; 175 struct mtx sc_events_mtx; 176 177 struct callout sc_callout; 178 }; 179 #define sc_name sc_geom->name 180 181 u_int g_mirror_ndisks(struct g_mirror_softc *sc, int state); 182 int g_mirror_destroy(struct g_mirror_softc *sc, boolean_t force); 183 int g_mirror_event_send(void *arg, int state, int flags); 184 struct g_mirror_metadata; 185 void g_mirror_fill_metadata(struct g_mirror_softc *sc, 186 struct g_mirror_disk *disk, struct g_mirror_metadata *md); 187 void g_mirror_update_metadata(struct g_mirror_disk *disk); 188 189 g_ctl_req_t g_mirror_config; 190 #endif /* _KERNEL */ 191 192 struct g_mirror_metadata { 193 char md_magic[16]; /* Magic value. */ 194 uint32_t md_version; /* Version number. */ 195 char md_name[16]; /* Mirror name. */ 196 uint32_t md_mid; /* Mirror unique ID. */ 197 uint32_t md_did; /* Disk unique ID. */ 198 uint8_t md_all; /* Number of disks in mirror. */ 199 uint32_t md_syncid; /* Synchronization ID. */ 200 uint8_t md_priority; /* Disk priority. */ 201 uint32_t md_slice; /* Slice size. */ 202 uint8_t md_balance; /* Balance type. */ 203 uint64_t md_mediasize; /* Size of the smallest 204 disk in mirror. */ 205 uint32_t md_sectorsize; /* Sector size. */ 206 uint64_t md_sync_offset; /* Synchronized offset. */ 207 uint64_t md_mflags; /* Additional mirror flags. */ 208 uint64_t md_dflags; /* Additional disk flags. */ 209 u_char md_hash[16]; /* MD5 hash. */ 210 }; 211 static __inline void 212 mirror_metadata_encode(struct g_mirror_metadata *md, u_char *data) 213 { 214 MD5_CTX ctx; 215 216 bcopy(md->md_magic, data, 16); 217 le32enc(data + 16, md->md_version); 218 bcopy(md->md_name, data + 20, 16); 219 le32enc(data + 36, md->md_mid); 220 le32enc(data + 40, md->md_did); 221 *(data + 44) = md->md_all; 222 le32enc(data + 45, md->md_syncid); 223 *(data + 49) = md->md_priority; 224 le32enc(data + 50, md->md_slice); 225 *(data + 54) = md->md_balance; 226 le64enc(data + 55, md->md_mediasize); 227 le32enc(data + 63, md->md_sectorsize); 228 le64enc(data + 67, md->md_sync_offset); 229 le64enc(data + 75, md->md_mflags); 230 le64enc(data + 83, md->md_dflags); 231 MD5Init(&ctx); 232 MD5Update(&ctx, data, 91); 233 MD5Final(md->md_hash, &ctx); 234 bcopy(md->md_hash, data + 91, 16); 235 } 236 static __inline int 237 mirror_metadata_decode(const u_char *data, struct g_mirror_metadata *md) 238 { 239 MD5_CTX ctx; 240 241 bcopy(data, md->md_magic, 16); 242 md->md_version = le32dec(data + 16); 243 bcopy(data + 20, md->md_name, 16); 244 md->md_mid = le32dec(data + 36); 245 md->md_did = le32dec(data + 40); 246 md->md_all = *(data + 44); 247 md->md_syncid = le32dec(data + 45); 248 md->md_priority = *(data + 49); 249 md->md_slice = le32dec(data + 50); 250 md->md_balance = *(data + 54); 251 md->md_mediasize = le64dec(data + 55); 252 md->md_sectorsize = le32dec(data + 63); 253 md->md_sync_offset = le64dec(data + 67); 254 md->md_mflags = le64dec(data + 75); 255 md->md_dflags = le64dec(data + 83); 256 bcopy(data + 91, md->md_hash, 16); 257 MD5Init(&ctx); 258 MD5Update(&ctx, data, 91); 259 MD5Final(md->md_hash, &ctx); 260 if (bcmp(md->md_hash, data + 91, 16) != 0) 261 return (EINVAL); 262 return (0); 263 } 264 265 static __inline const char * 266 balance_name(u_int balance) 267 { 268 static const char *algorithms[] = { 269 [G_MIRROR_BALANCE_NONE] = "none", 270 [G_MIRROR_BALANCE_ROUND_ROBIN] = "round-robin", 271 [G_MIRROR_BALANCE_LOAD] = "load", 272 [G_MIRROR_BALANCE_SPLIT] = "split", 273 [G_MIRROR_BALANCE_MAX + 1] = "unknown" 274 }; 275 276 if (balance > G_MIRROR_BALANCE_MAX) 277 balance = G_MIRROR_BALANCE_MAX + 1; 278 279 return (algorithms[balance]); 280 } 281 282 static __inline int 283 balance_id(const char *name) 284 { 285 static const char *algorithms[] = { 286 [G_MIRROR_BALANCE_NONE] = "none", 287 [G_MIRROR_BALANCE_ROUND_ROBIN] = "round-robin", 288 [G_MIRROR_BALANCE_LOAD] = "load", 289 [G_MIRROR_BALANCE_SPLIT] = "split" 290 }; 291 int n; 292 293 for (n = G_MIRROR_BALANCE_MIN; n <= G_MIRROR_BALANCE_MAX; n++) { 294 if (strcmp(name, algorithms[n]) == 0) 295 return (n); 296 } 297 return (-1); 298 } 299 300 static __inline void 301 mirror_metadata_dump(const struct g_mirror_metadata *md) 302 { 303 static const char hex[] = "0123456789abcdef"; 304 char hash[16 * 2 + 1]; 305 u_int i; 306 307 printf(" magic: %s\n", md->md_magic); 308 printf(" version: %u\n", (u_int)md->md_version); 309 printf(" name: %s\n", md->md_name); 310 printf(" mid: %u\n", (u_int)md->md_mid); 311 printf(" did: %u\n", (u_int)md->md_did); 312 printf(" all: %u\n", (u_int)md->md_all); 313 printf(" syncid: %u\n", (u_int)md->md_syncid); 314 printf(" priority: %u\n", (u_int)md->md_priority); 315 printf(" slice: %u\n", (u_int)md->md_slice); 316 printf(" balance: %s\n", balance_name((u_int)md->md_balance)); 317 printf(" mediasize: %jd\n", (intmax_t)md->md_mediasize); 318 printf("sectorsize: %u\n", (u_int)md->md_sectorsize); 319 printf("syncoffset: %jd\n", (intmax_t)md->md_sync_offset); 320 printf(" mflags:"); 321 if (md->md_mflags == 0) 322 printf(" NONE"); 323 else { 324 if ((md->md_mflags & G_MIRROR_DEVICE_FLAG_NOAUTOSYNC) != 0) 325 printf(" NOAUTOSYNC"); 326 } 327 printf("\n"); 328 printf(" dflags:"); 329 if (md->md_mflags == 0) 330 printf(" NONE"); 331 else { 332 if ((md->md_dflags & G_MIRROR_DISK_FLAG_DIRTY) != 0) 333 printf(" DIRTY"); 334 if ((md->md_dflags & G_MIRROR_DISK_FLAG_SYNCHRONIZING) != 0) 335 printf(" SYNCHRONIZING"); 336 if ((md->md_dflags & G_MIRROR_DISK_FLAG_FORCE_SYNC) != 0) 337 printf(" FORCE_SYNC"); 338 if ((md->md_dflags & G_MIRROR_DISK_FLAG_INACTIVE) != 0) 339 printf(" INACTIVE"); 340 } 341 printf("\n"); 342 bzero(hash, sizeof(hash)); 343 for (i = 0; i < 16; i++) { 344 hash[i * 2] = hex[md->md_hash[i] >> 4]; 345 hash[i * 2 + 1] = hex[md->md_hash[i] & 0x0f]; 346 } 347 printf(" MD5 hash: %s\n", hash); 348 } 349 #endif /* !_G_MIRROR_H_ */ 350