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 /* 23 * Copyright 2007 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #ifndef _ZIO_H 28 #define _ZIO_H 29 30 #pragma ident "%Z%%M% %I% %E% SMI" 31 32 #include <sys/zfs_context.h> 33 #include <sys/spa.h> 34 #include <sys/txg.h> 35 #include <sys/avl.h> 36 #include <sys/dkio.h> 37 #include <sys/fs/zfs.h> 38 #include <sys/zio_impl.h> 39 40 #ifdef __cplusplus 41 extern "C" { 42 #endif 43 44 #define ZBT_MAGIC 0x210da7ab10c7a11ULL /* zio data bloc tail */ 45 46 typedef struct zio_block_tail { 47 uint64_t zbt_magic; /* for validation, endianness */ 48 zio_cksum_t zbt_cksum; /* 256-bit checksum */ 49 } zio_block_tail_t; 50 51 /* 52 * Gang block headers are self-checksumming and contain an array 53 * of block pointers. 54 */ 55 #define SPA_GANGBLOCKSIZE SPA_MINBLOCKSIZE 56 #define SPA_GBH_NBLKPTRS ((SPA_GANGBLOCKSIZE - \ 57 sizeof (zio_block_tail_t)) / sizeof (blkptr_t)) 58 #define SPA_GBH_FILLER ((SPA_GANGBLOCKSIZE - \ 59 sizeof (zio_block_tail_t) - \ 60 (SPA_GBH_NBLKPTRS * sizeof (blkptr_t))) /\ 61 sizeof (uint64_t)) 62 63 #define ZIO_GET_IOSIZE(zio) \ 64 (BP_IS_GANG((zio)->io_bp) ? \ 65 SPA_GANGBLOCKSIZE : BP_GET_PSIZE((zio)->io_bp)) 66 67 typedef struct zio_gbh { 68 blkptr_t zg_blkptr[SPA_GBH_NBLKPTRS]; 69 uint64_t zg_filler[SPA_GBH_FILLER]; 70 zio_block_tail_t zg_tail; 71 } zio_gbh_phys_t; 72 73 enum zio_checksum { 74 ZIO_CHECKSUM_INHERIT = 0, 75 ZIO_CHECKSUM_ON, 76 ZIO_CHECKSUM_OFF, 77 ZIO_CHECKSUM_LABEL, 78 ZIO_CHECKSUM_GANG_HEADER, 79 ZIO_CHECKSUM_ZILOG, 80 ZIO_CHECKSUM_FLETCHER_2, 81 ZIO_CHECKSUM_FLETCHER_4, 82 ZIO_CHECKSUM_SHA256, 83 ZIO_CHECKSUM_FUNCTIONS 84 }; 85 86 #define ZIO_CHECKSUM_ON_VALUE ZIO_CHECKSUM_FLETCHER_2 87 #define ZIO_CHECKSUM_DEFAULT ZIO_CHECKSUM_ON 88 89 enum zio_compress { 90 ZIO_COMPRESS_INHERIT = 0, 91 ZIO_COMPRESS_ON, 92 ZIO_COMPRESS_OFF, 93 ZIO_COMPRESS_LZJB, 94 ZIO_COMPRESS_EMPTY, 95 ZIO_COMPRESS_GZIP_1, 96 ZIO_COMPRESS_GZIP_2, 97 ZIO_COMPRESS_GZIP_3, 98 ZIO_COMPRESS_GZIP_4, 99 ZIO_COMPRESS_GZIP_5, 100 ZIO_COMPRESS_GZIP_6, 101 ZIO_COMPRESS_GZIP_7, 102 ZIO_COMPRESS_GZIP_8, 103 ZIO_COMPRESS_GZIP_9, 104 ZIO_COMPRESS_FUNCTIONS 105 }; 106 107 #define ZIO_COMPRESS_ON_VALUE ZIO_COMPRESS_LZJB 108 #define ZIO_COMPRESS_DEFAULT ZIO_COMPRESS_OFF 109 110 #define ZIO_FAILURE_MODE_WAIT 0 111 #define ZIO_FAILURE_MODE_CONTINUE 1 112 #define ZIO_FAILURE_MODE_PANIC 2 113 114 #define ZIO_PRIORITY_NOW (zio_priority_table[0]) 115 #define ZIO_PRIORITY_SYNC_READ (zio_priority_table[1]) 116 #define ZIO_PRIORITY_SYNC_WRITE (zio_priority_table[2]) 117 #define ZIO_PRIORITY_ASYNC_READ (zio_priority_table[3]) 118 #define ZIO_PRIORITY_ASYNC_WRITE (zio_priority_table[4]) 119 #define ZIO_PRIORITY_FREE (zio_priority_table[5]) 120 #define ZIO_PRIORITY_CACHE_FILL (zio_priority_table[6]) 121 #define ZIO_PRIORITY_LOG_WRITE (zio_priority_table[7]) 122 #define ZIO_PRIORITY_RESILVER (zio_priority_table[8]) 123 #define ZIO_PRIORITY_SCRUB (zio_priority_table[9]) 124 #define ZIO_PRIORITY_TABLE_SIZE 10 125 126 #define ZIO_FLAG_MUSTSUCCEED 0x00000 127 #define ZIO_FLAG_CANFAIL 0x00001 128 #define ZIO_FLAG_FAILFAST 0x00002 129 #define ZIO_FLAG_CONFIG_HELD 0x00004 130 #define ZIO_FLAG_CONFIG_GRABBED 0x00008 131 132 #define ZIO_FLAG_DONT_CACHE 0x00010 133 #define ZIO_FLAG_DONT_QUEUE 0x00020 134 #define ZIO_FLAG_DONT_PROPAGATE 0x00040 135 #define ZIO_FLAG_DONT_RETRY 0x00080 136 137 #define ZIO_FLAG_PHYSICAL 0x00100 138 #define ZIO_FLAG_IO_BYPASS 0x00200 139 #define ZIO_FLAG_IO_REPAIR 0x00400 140 #define ZIO_FLAG_SPECULATIVE 0x00800 141 142 #define ZIO_FLAG_RESILVER 0x01000 143 #define ZIO_FLAG_SCRUB 0x02000 144 #define ZIO_FLAG_SCRUB_THREAD 0x04000 145 #define ZIO_FLAG_SUBBLOCK 0x08000 146 147 #define ZIO_FLAG_NOBOOKMARK 0x10000 148 #define ZIO_FLAG_USER 0x20000 149 150 #define ZIO_FLAG_METADATA 0x40000 151 #define ZIO_FLAG_WRITE_RETRY 0x80000 152 153 #define ZIO_FLAG_GANG_INHERIT \ 154 (ZIO_FLAG_CANFAIL | \ 155 ZIO_FLAG_FAILFAST | \ 156 ZIO_FLAG_CONFIG_HELD | \ 157 ZIO_FLAG_DONT_RETRY | \ 158 ZIO_FLAG_IO_REPAIR | \ 159 ZIO_FLAG_SPECULATIVE | \ 160 ZIO_FLAG_RESILVER | \ 161 ZIO_FLAG_SCRUB | \ 162 ZIO_FLAG_SCRUB_THREAD) 163 164 #define ZIO_FLAG_VDEV_INHERIT \ 165 (ZIO_FLAG_GANG_INHERIT | \ 166 ZIO_FLAG_DONT_CACHE | \ 167 ZIO_FLAG_PHYSICAL) 168 169 /* 170 * We'll take the unused errno 'EBADE' (from the Convergent graveyard) 171 * to indicate checksum errors. 172 */ 173 #define ECKSUM EBADE 174 175 typedef struct zio zio_t; 176 typedef void zio_done_func_t(zio_t *zio); 177 178 extern uint8_t zio_priority_table[ZIO_PRIORITY_TABLE_SIZE]; 179 extern char *zio_type_name[ZIO_TYPES]; 180 181 /* 182 * A bookmark is a four-tuple <objset, object, level, blkid> that uniquely 183 * identifies any block in the pool. By convention, the meta-objset (MOS) 184 * is objset 0, the meta-dnode is object 0, the root block (osphys_t) is 185 * level -1 of the meta-dnode, and intent log blocks (which are chained 186 * off the root block) have blkid == sequence number. In summary: 187 * 188 * mos is objset 0 189 * meta-dnode is object 0 190 * root block is <objset, 0, -1, 0> 191 * intent log is <objset, 0, -1, ZIL sequence number> 192 * 193 * Note: this structure is called a bookmark because its first purpose was 194 * to remember where to resume a pool-wide traverse. The absolute ordering 195 * for block visitation during traversal is defined in compare_bookmark(). 196 * 197 * Note: this structure is passed between userland and the kernel. 198 * Therefore it must not change size or alignment between 32/64 bit 199 * compilation options. 200 */ 201 typedef struct zbookmark { 202 uint64_t zb_objset; 203 uint64_t zb_object; 204 int64_t zb_level; 205 uint64_t zb_blkid; 206 } zbookmark_t; 207 208 struct zio { 209 /* Core information about this I/O */ 210 zio_t *io_parent; 211 zio_t *io_root; 212 spa_t *io_spa; 213 zbookmark_t io_bookmark; 214 enum zio_checksum io_checksum; 215 enum zio_compress io_compress; 216 int io_ndvas; 217 uint64_t io_txg; 218 blkptr_t *io_bp; 219 blkptr_t io_bp_copy; 220 zio_t *io_child; 221 zio_t *io_sibling_prev; 222 zio_t *io_sibling_next; 223 zio_transform_t *io_transform_stack; 224 zio_t *io_logical; 225 list_node_t zio_link_node; 226 227 /* Callback info */ 228 zio_done_func_t *io_ready; 229 zio_done_func_t *io_done; 230 void *io_private; 231 blkptr_t io_bp_orig; 232 233 /* Data represented by this I/O */ 234 void *io_data; 235 uint64_t io_size; 236 237 /* Stuff for the vdev stack */ 238 vdev_t *io_vd; 239 void *io_vsd; 240 uint64_t io_offset; 241 uint64_t io_deadline; 242 uint64_t io_timestamp; 243 avl_node_t io_offset_node; 244 avl_node_t io_deadline_node; 245 avl_tree_t *io_vdev_tree; 246 zio_t *io_delegate_list; 247 zio_t *io_delegate_next; 248 249 /* Internal pipeline state */ 250 int io_flags; 251 int io_orig_flags; 252 enum zio_type io_type; 253 enum zio_stage io_stage; 254 enum zio_stage io_orig_stage; 255 uint8_t io_stalled; 256 uint8_t io_priority; 257 struct dk_callback io_dk_callback; 258 int io_cmd; 259 int io_retries; 260 int io_error; 261 uint32_t io_numerrors; 262 uint32_t io_pipeline; 263 uint32_t io_orig_pipeline; 264 uint32_t io_async_stages; 265 uint64_t io_children_notready; 266 uint64_t io_children_notdone; 267 void *io_waiter; 268 kmutex_t io_lock; 269 kcondvar_t io_cv; 270 271 /* FMA state */ 272 uint64_t io_ena; 273 }; 274 275 extern zio_t *zio_null(zio_t *pio, spa_t *spa, 276 zio_done_func_t *done, void *private, int flags); 277 278 extern zio_t *zio_root(spa_t *spa, 279 zio_done_func_t *done, void *private, int flags); 280 281 extern zio_t *zio_read(zio_t *pio, spa_t *spa, blkptr_t *bp, void *data, 282 uint64_t size, zio_done_func_t *done, void *private, 283 int priority, int flags, zbookmark_t *zb); 284 285 extern zio_t *zio_write(zio_t *pio, spa_t *spa, int checksum, int compress, 286 int ncopies, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 287 zio_done_func_t *ready, zio_done_func_t *done, void *private, int priority, 288 int flags, zbookmark_t *zb); 289 290 extern zio_t *zio_rewrite(zio_t *pio, spa_t *spa, int checksum, 291 uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 292 zio_done_func_t *done, void *private, int priority, int flags, 293 zbookmark_t *zb); 294 295 extern zio_t *zio_free(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 296 zio_done_func_t *done, void *private); 297 298 extern zio_t *zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 299 zio_done_func_t *done, void *private); 300 301 extern zio_t *zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, 302 zio_done_func_t *done, void *private, int priority, int flags); 303 304 extern zio_t *zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 305 uint64_t size, void *data, int checksum, 306 zio_done_func_t *done, void *private, int priority, int flags); 307 308 extern zio_t *zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 309 uint64_t size, void *data, int checksum, 310 zio_done_func_t *done, void *private, int priority, int flags); 311 312 extern int zio_alloc_blk(spa_t *spa, uint64_t size, blkptr_t *new_bp, 313 blkptr_t *old_bp, uint64_t txg); 314 extern void zio_free_blk(spa_t *spa, blkptr_t *bp, uint64_t txg); 315 extern void zio_flush_vdev(spa_t *spa, uint64_t vdev, zio_t **zio); 316 317 extern int zio_wait(zio_t *zio); 318 extern void zio_nowait(zio_t *zio); 319 320 extern void *zio_buf_alloc(size_t size); 321 extern void zio_buf_free(void *buf, size_t size); 322 extern void *zio_data_buf_alloc(size_t size); 323 extern void zio_data_buf_free(void *buf, size_t size); 324 325 /* 326 * Move an I/O to the next stage of the pipeline and execute that stage. 327 * There's no locking on io_stage because there's no legitimate way for 328 * multiple threads to be attempting to process the same I/O. 329 */ 330 extern void zio_next_stage(zio_t *zio); 331 extern void zio_next_stage_async(zio_t *zio); 332 extern void zio_resubmit_stage_async(void *); 333 extern void zio_wait_children_done(zio_t *zio); 334 335 /* 336 * Delegate I/O to a child vdev. 337 */ 338 extern zio_t *zio_vdev_child_io(zio_t *zio, blkptr_t *bp, vdev_t *vd, 339 uint64_t offset, void *data, uint64_t size, int type, int priority, 340 int flags, zio_done_func_t *done, void *private); 341 342 extern void zio_vdev_io_bypass(zio_t *zio); 343 extern void zio_vdev_io_reissue(zio_t *zio); 344 extern void zio_vdev_io_redone(zio_t *zio); 345 346 extern void zio_checksum_verified(zio_t *zio); 347 extern void zio_set_gang_verifier(zio_t *zio, zio_cksum_t *zcp); 348 349 extern uint8_t zio_checksum_select(uint8_t child, uint8_t parent); 350 extern uint8_t zio_compress_select(uint8_t child, uint8_t parent); 351 352 extern boolean_t zio_should_retry(zio_t *zio); 353 extern int zio_vdev_resume_io(spa_t *); 354 355 /* 356 * Initial setup and teardown. 357 */ 358 extern void zio_init(void); 359 extern void zio_fini(void); 360 361 /* 362 * Fault injection 363 */ 364 struct zinject_record; 365 extern uint32_t zio_injection_enabled; 366 extern int zio_inject_fault(char *name, int flags, int *id, 367 struct zinject_record *record); 368 extern int zio_inject_list_next(int *id, char *name, size_t buflen, 369 struct zinject_record *record); 370 extern int zio_clear_fault(int id); 371 extern int zio_handle_fault_injection(zio_t *zio, int error); 372 extern int zio_handle_device_injection(vdev_t *vd, int error); 373 374 #ifdef __cplusplus 375 } 376 #endif 377 378 #endif /* _ZIO_H */ 379