1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 /* 22 * Copyright 2007 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _ZIO_H 27 #define _ZIO_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #include <sys/zfs_context.h> 32 #include <sys/spa.h> 33 #include <sys/txg.h> 34 #include <sys/avl.h> 35 #include <sys/dkio.h> 36 #include <sys/fs/zfs.h> 37 #include <sys/zio_impl.h> 38 39 #ifdef __cplusplus 40 extern "C" { 41 #endif 42 43 #define ZBT_MAGIC 0x210da7ab10c7a11ULL /* zio data bloc tail */ 44 45 typedef struct zio_block_tail { 46 uint64_t zbt_magic; /* for validation, endianness */ 47 zio_cksum_t zbt_cksum; /* 256-bit checksum */ 48 } zio_block_tail_t; 49 50 /* 51 * Gang block headers are self-checksumming and contain an array 52 * of block pointers. 53 */ 54 #define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE 55 #define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \ 56 sizeof (zio_block_tail_t)) / sizeof (blkptr_t)) 57 #define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \ 58 sizeof (zio_block_tail_t) - \ 59 (SPA_GBH_NBLKPTRS * sizeof (blkptr_t))) /\ 60 sizeof (uint64_t)) 61 62 #define ZIO_GET_IOSIZE(zio) \ 63 (BP_IS_GANG((zio)->io_bp) ? \ 64 SPA_GANGBLOCKSIZE : BP_GET_PSIZE((zio)->io_bp)) 65 66 typedef struct zio_gbh { 67 blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS]; 68 uint64_t zg_filler[SPA_GBH_FILLER]; 69 zio_block_tail_t zg_tail; 70 } zio_gbh_phys_t; 71 72 enum zio_checksum { 73 ZIO_CHECKSUM_INHERIT = 0, 74 ZIO_CHECKSUM_ON, 75 ZIO_CHECKSUM_OFF, 76 ZIO_CHECKSUM_LABEL, 77 ZIO_CHECKSUM_GANG_HEADER, 78 ZIO_CHECKSUM_ZILOG, 79 ZIO_CHECKSUM_FLETCHER_2, 80 ZIO_CHECKSUM_FLETCHER_4, 81 ZIO_CHECKSUM_SHA256, 82 ZIO_CHECKSUM_FUNCTIONS 83 }; 84 85 #define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_2 86 #define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON 87 88 enum zio_compress { 89 ZIO_COMPRESS_INHERIT = 0, 90 ZIO_COMPRESS_ON, 91 ZIO_COMPRESS_OFF, 92 ZIO_COMPRESS_LZJB, 93 ZIO_COMPRESS_EMPTY, 94 ZIO_COMPRESS_FUNCTIONS 95 }; 96 97 #define ZIO_COMPRESS_ON_VALUE ZIO_COMPRESS_LZJB 98 #define ZIO_COMPRESS_DEFAULT ZIO_COMPRESS_OFF 99 100 #define ZIO_PRIORITY_NOW (zio_priority_table[0]) 101 #define ZIO_PRIORITY_SYNC_READ (zio_priority_table[1]) 102 #define ZIO_PRIORITY_SYNC_WRITE (zio_priority_table[2]) 103 #define ZIO_PRIORITY_ASYNC_READ (zio_priority_table[3]) 104 #define ZIO_PRIORITY_ASYNC_WRITE (zio_priority_table[4]) 105 #define ZIO_PRIORITY_FREE (zio_priority_table[5]) 106 #define ZIO_PRIORITY_CACHE_FILL (zio_priority_table[6]) 107 #define ZIO_PRIORITY_LOG_WRITE (zio_priority_table[7]) 108 #define ZIO_PRIORITY_RESILVER (zio_priority_table[8]) 109 #define ZIO_PRIORITY_SCRUB (zio_priority_table[9]) 110 #define ZIO_PRIORITY_TABLE_SIZE 10 111 112 #define ZIO_FLAG_MUSTSUCCEED 0x00000 113 #define ZIO_FLAG_CANFAIL 0x00001 114 #define ZIO_FLAG_FAILFAST 0x00002 115 #define ZIO_FLAG_CONFIG_HELD 0x00004 116 #define ZIO_FLAG_CONFIG_GRABBED 0x00008 117 118 #define ZIO_FLAG_DONT_CACHE 0x00010 119 #define ZIO_FLAG_DONT_QUEUE 0x00020 120 #define ZIO_FLAG_DONT_PROPAGATE 0x00040 121 #define ZIO_FLAG_DONT_RETRY 0x00080 122 123 #define ZIO_FLAG_PHYSICAL 0x00100 124 #define ZIO_FLAG_IO_BYPASS 0x00200 125 #define ZIO_FLAG_IO_REPAIR 0x00400 126 #define ZIO_FLAG_SPECULATIVE 0x00800 127 128 #define ZIO_FLAG_RESILVER 0x01000 129 #define ZIO_FLAG_SCRUB 0x02000 130 #define ZIO_FLAG_SCRUB_THREAD 0x04000 131 #define ZIO_FLAG_SUBBLOCK 0x08000 132 133 #define ZIO_FLAG_NOBOOKMARK 0x10000 134 #define ZIO_FLAG_USER 0x20000 135 136 #define ZIO_FLAG_METADATA 0x40000 137 138 #define ZIO_FLAG_GANG_INHERIT \ 139 (ZIO_FLAG_CANFAIL | \ 140 ZIO_FLAG_FAILFAST | \ 141 ZIO_FLAG_CONFIG_HELD | \ 142 ZIO_FLAG_DONT_RETRY | \ 143 ZIO_FLAG_IO_REPAIR | \ 144 ZIO_FLAG_SPECULATIVE | \ 145 ZIO_FLAG_RESILVER | \ 146 ZIO_FLAG_SCRUB | \ 147 ZIO_FLAG_SCRUB_THREAD) 148 149 #define ZIO_FLAG_VDEV_INHERIT \ 150 (ZIO_FLAG_GANG_INHERIT | \ 151 ZIO_FLAG_DONT_CACHE | \ 152 ZIO_FLAG_PHYSICAL) 153 154 /* 155 * We'll take the unused errno 'EBADE' (from the Convergent graveyard) 156 * to indicate checksum errors. 157 */ 158 #define ECKSUM EBADE 159 160 typedef struct zio zio_t; 161 typedef void zio_done_func_t(zio_t *zio); 162 163 extern uint8_t zio_priority_table[ZIO_PRIORITY_TABLE_SIZE]; 164 extern char *zio_type_name[ZIO_TYPES]; 165 166 /* 167 * A bookmark is a four-tuple <objset, object, level, blkid> that uniquely 168 * identifies any block in the pool. By convention, the meta-objset (MOS) 169 * is objset 0, the meta-dnode is object 0, the root block (osphys_t) is 170 * level -1 of the meta-dnode, and intent log blocks (which are chained 171 * off the root block) have blkid == sequence number. In summary: 172 * 173 * mos is objset 0 174 * meta-dnode is object 0 175 * root block is <objset, 0, -1, 0> 176 * intent log is <objset, 0, -1, ZIL sequence number> 177 * 178 * Note: this structure is called a bookmark because its first purpose was 179 * to remember where to resume a pool-wide traverse. The absolute ordering 180 * for block visitation during traversal is defined in compare_bookmark(). 181 * 182 * Note: this structure is passed between userland and the kernel. 183 * Therefore it must not change size or alignment between 32/64 bit 184 * compilation options. 185 */ 186 typedef struct zbookmark { 187 uint64_t zb_objset; 188 uint64_t zb_object; 189 int64_t zb_level; 190 uint64_t zb_blkid; 191 } zbookmark_t; 192 193 struct zio { 194 /* Core information about this I/O */ 195 zio_t *io_parent; 196 zio_t *io_root; 197 spa_t *io_spa; 198 zbookmark_t io_bookmark; 199 enum zio_checksum io_checksum; 200 enum zio_compress io_compress; 201 int io_ndvas; 202 uint64_t io_txg; 203 blkptr_t *io_bp; 204 blkptr_t io_bp_copy; 205 zio_t *io_child; 206 zio_t *io_sibling_prev; 207 zio_t *io_sibling_next; 208 zio_transform_t *io_transform_stack; 209 zio_t *io_logical; 210 211 /* Callback info */ 212 zio_done_func_t *io_ready; 213 zio_done_func_t *io_done; 214 void *io_private; 215 blkptr_t io_bp_orig; 216 217 /* Data represented by this I/O */ 218 void *io_data; 219 uint64_t io_size; 220 221 /* Stuff for the vdev stack */ 222 vdev_t *io_vd; 223 void *io_vsd; 224 uint64_t io_offset; 225 uint64_t io_deadline; 226 uint64_t io_timestamp; 227 avl_node_t io_offset_node; 228 avl_node_t io_deadline_node; 229 avl_tree_t *io_vdev_tree; 230 zio_t *io_delegate_list; 231 zio_t *io_delegate_next; 232 233 /* Internal pipeline state */ 234 int io_flags; 235 enum zio_type io_type; 236 enum zio_stage io_stage; 237 uint8_t io_stalled; 238 uint8_t io_priority; 239 struct dk_callback io_dk_callback; 240 int io_cmd; 241 int io_retries; 242 int io_error; 243 uint32_t io_numerrors; 244 uint32_t io_pipeline; 245 uint32_t io_async_stages; 246 uint64_t io_children_notready; 247 uint64_t io_children_notdone; 248 void *io_waiter; 249 kmutex_t io_lock; 250 kcondvar_t io_cv; 251 252 /* FMA state */ 253 uint64_t io_ena; 254 }; 255 256 extern zio_t *zio_null(zio_t *pio, spa_t *spa, 257 zio_done_func_t *done, void *private, int flags); 258 259 extern zio_t *zio_root(spa_t *spa, 260 zio_done_func_t *done, void *private, int flags); 261 262 extern zio_t *zio_read(zio_t *pio, spa_t *spa, blkptr_t *bp, void *data, 263 uint64_t size, zio_done_func_t *done, void *private, 264 int priority, int flags, zbookmark_t *zb); 265 266 extern zio_t *zio_write(zio_t *pio, spa_t *spa, int checksum, int compress, 267 int ncopies, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 268 zio_done_func_t *ready, zio_done_func_t *done, void *private, int priority, 269 int flags, zbookmark_t *zb); 270 271 extern zio_t *zio_rewrite(zio_t *pio, spa_t *spa, int checksum, 272 uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 273 zio_done_func_t *done, void *private, int priority, int flags, 274 zbookmark_t *zb); 275 276 extern zio_t *zio_free(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 277 zio_done_func_t *done, void *private); 278 279 extern zio_t *zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 280 zio_done_func_t *done, void *private); 281 282 extern zio_t *zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, 283 zio_done_func_t *done, void *private, int priority, int flags); 284 285 extern zio_t *zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 286 uint64_t size, void *data, int checksum, 287 zio_done_func_t *done, void *private, int priority, int flags); 288 289 extern zio_t *zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 290 uint64_t size, void *data, int checksum, 291 zio_done_func_t *done, void *private, int priority, int flags); 292 293 extern int zio_alloc_blk(spa_t *spa, uint64_t size, blkptr_t *new_bp, 294 blkptr_t *old_bp, uint64_t txg); 295 extern void zio_free_blk(spa_t *spa, blkptr_t *bp, uint64_t txg); 296 297 extern int zio_wait(zio_t *zio); 298 extern void zio_nowait(zio_t *zio); 299 300 extern void *zio_buf_alloc(size_t size); 301 extern void zio_buf_free(void *buf, size_t size); 302 extern void *zio_data_buf_alloc(size_t size); 303 extern void zio_data_buf_free(void *buf, size_t size); 304 305 /* 306 * Move an I/O to the next stage of the pipeline and execute that stage. 307 * There's no locking on io_stage because there's no legitimate way for 308 * multiple threads to be attempting to process the same I/O. 309 */ 310 extern void zio_next_stage(zio_t *zio); 311 extern void zio_next_stage_async(zio_t *zio); 312 extern void zio_wait_children_done(zio_t *zio); 313 314 /* 315 * Delegate I/O to a child vdev. 316 */ 317 extern zio_t *zio_vdev_child_io(zio_t *zio, blkptr_t *bp, vdev_t *vd, 318 uint64_t offset, void *data, uint64_t size, int type, int priority, 319 int flags, zio_done_func_t *done, void *private); 320 321 extern void zio_vdev_io_bypass(zio_t *zio); 322 extern void zio_vdev_io_reissue(zio_t *zio); 323 extern void zio_vdev_io_redone(zio_t *zio); 324 325 extern void zio_checksum_verified(zio_t *zio); 326 extern void zio_set_gang_verifier(zio_t *zio, zio_cksum_t *zcp); 327 328 extern uint8_t zio_checksum_select(uint8_t child, uint8_t parent); 329 extern uint8_t zio_compress_select(uint8_t child, uint8_t parent); 330 331 boolean_t zio_should_retry(zio_t *zio); 332 333 /* 334 * Initial setup and teardown. 335 */ 336 extern void zio_init(void); 337 extern void zio_fini(void); 338 339 /* 340 * Fault injection 341 */ 342 struct zinject_record; 343 extern uint32_t zio_injection_enabled; 344 extern int zio_inject_fault(char *name, int flags, int *id, 345 struct zinject_record *record); 346 extern int zio_inject_list_next(int *id, char *name, size_t buflen, 347 struct zinject_record *record); 348 extern int zio_clear_fault(int id); 349 extern int zio_handle_fault_injection(zio_t *zio, int error); 350 extern int zio_handle_device_injection(vdev_t *vd, int error); 351 352 #ifdef __cplusplus 353 } 354 #endif 355 356 #endif /* _ZIO_H */ 357