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_GANG_INHERIT \ 137 (ZIO_FLAG_CANFAIL | \ 138 ZIO_FLAG_FAILFAST | \ 139 ZIO_FLAG_CONFIG_HELD | \ 140 ZIO_FLAG_DONT_RETRY | \ 141 ZIO_FLAG_IO_REPAIR | \ 142 ZIO_FLAG_SPECULATIVE | \ 143 ZIO_FLAG_RESILVER | \ 144 ZIO_FLAG_SCRUB | \ 145 ZIO_FLAG_SCRUB_THREAD) 146 147 #define ZIO_FLAG_VDEV_INHERIT \ 148 (ZIO_FLAG_GANG_INHERIT | \ 149 ZIO_FLAG_DONT_CACHE | \ 150 ZIO_FLAG_PHYSICAL) 151 152 /* 153 * We'll take the unused errno 'EBADE' (from the Convergent graveyard) 154 * to indicate checksum errors. 155 */ 156 #define ECKSUM EBADE 157 158 typedef struct zio zio_t; 159 typedef void zio_done_func_t(zio_t *zio); 160 161 extern uint8_t zio_priority_table[ZIO_PRIORITY_TABLE_SIZE]; 162 extern char *zio_type_name[ZIO_TYPES]; 163 164 /* 165 * A bookmark is a four-tuple <objset, object, level, blkid> that uniquely 166 * identifies any block in the pool. By convention, the meta-objset (MOS) 167 * is objset 0, the meta-dnode is object 0, the root block (osphys_t) is 168 * level -1 of the meta-dnode, and intent log blocks (which are chained 169 * off the root block) have blkid == sequence number. In summary: 170 * 171 * mos is objset 0 172 * meta-dnode is object 0 173 * root block is <objset, 0, -1, 0> 174 * intent log is <objset, 0, -1, ZIL sequence number> 175 * 176 * Note: this structure is called a bookmark because its first purpose was 177 * to remember where to resume a pool-wide traverse. The absolute ordering 178 * for block visitation during traversal is defined in compare_bookmark(). 179 * 180 * Note: this structure is passed between userland and the kernel. 181 * Therefore it must not change size or alignment between 32/64 bit 182 * compilation options. 183 */ 184 typedef struct zbookmark { 185 uint64_t zb_objset; 186 uint64_t zb_object; 187 int64_t zb_level; 188 uint64_t zb_blkid; 189 } zbookmark_t; 190 191 struct zio { 192 /* Core information about this I/O */ 193 zio_t *io_parent; 194 zio_t *io_root; 195 spa_t *io_spa; 196 zbookmark_t io_bookmark; 197 enum zio_checksum io_checksum; 198 enum zio_compress io_compress; 199 int io_ndvas; 200 uint64_t io_txg; 201 blkptr_t *io_bp; 202 blkptr_t io_bp_copy; 203 zio_t *io_child; 204 zio_t *io_sibling_prev; 205 zio_t *io_sibling_next; 206 zio_transform_t *io_transform_stack; 207 zio_t *io_logical; 208 209 /* Callback info */ 210 zio_done_func_t *io_ready; 211 zio_done_func_t *io_done; 212 void *io_private; 213 blkptr_t io_bp_orig; 214 215 /* Data represented by this I/O */ 216 void *io_data; 217 uint64_t io_size; 218 219 /* Stuff for the vdev stack */ 220 vdev_t *io_vd; 221 void *io_vsd; 222 uint64_t io_offset; 223 uint64_t io_deadline; 224 uint64_t io_timestamp; 225 avl_node_t io_offset_node; 226 avl_node_t io_deadline_node; 227 avl_tree_t *io_vdev_tree; 228 zio_t *io_delegate_list; 229 zio_t *io_delegate_next; 230 231 /* Internal pipeline state */ 232 int io_flags; 233 enum zio_type io_type; 234 enum zio_stage io_stage; 235 uint8_t io_stalled; 236 uint8_t io_priority; 237 struct dk_callback io_dk_callback; 238 int io_cmd; 239 int io_retries; 240 int io_error; 241 uint32_t io_numerrors; 242 uint32_t io_pipeline; 243 uint32_t io_async_stages; 244 uint64_t io_children_notready; 245 uint64_t io_children_notdone; 246 void *io_waiter; 247 kmutex_t io_lock; 248 kcondvar_t io_cv; 249 250 /* FMA state */ 251 uint64_t io_ena; 252 }; 253 254 extern zio_t *zio_null(zio_t *pio, spa_t *spa, 255 zio_done_func_t *done, void *private, int flags); 256 257 extern zio_t *zio_root(spa_t *spa, 258 zio_done_func_t *done, void *private, int flags); 259 260 extern zio_t *zio_read(zio_t *pio, spa_t *spa, blkptr_t *bp, void *data, 261 uint64_t size, zio_done_func_t *done, void *private, 262 int priority, int flags, zbookmark_t *zb); 263 264 extern zio_t *zio_write(zio_t *pio, spa_t *spa, int checksum, int compress, 265 int ncopies, uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 266 zio_done_func_t *ready, zio_done_func_t *done, void *private, int priority, 267 int flags, zbookmark_t *zb); 268 269 extern zio_t *zio_rewrite(zio_t *pio, spa_t *spa, int checksum, 270 uint64_t txg, blkptr_t *bp, void *data, uint64_t size, 271 zio_done_func_t *done, void *private, int priority, int flags, 272 zbookmark_t *zb); 273 274 extern zio_t *zio_free(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 275 zio_done_func_t *done, void *private); 276 277 extern zio_t *zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, 278 zio_done_func_t *done, void *private); 279 280 extern zio_t *zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, 281 zio_done_func_t *done, void *private, int priority, int flags); 282 283 extern zio_t *zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 284 uint64_t size, void *data, int checksum, 285 zio_done_func_t *done, void *private, int priority, int flags); 286 287 extern zio_t *zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, 288 uint64_t size, void *data, int checksum, 289 zio_done_func_t *done, void *private, int priority, int flags); 290 291 extern int zio_alloc_blk(spa_t *spa, uint64_t size, blkptr_t *new_bp, 292 blkptr_t *old_bp, uint64_t txg); 293 extern void zio_free_blk(spa_t *spa, blkptr_t *bp, uint64_t txg); 294 295 extern int zio_wait(zio_t *zio); 296 extern void zio_nowait(zio_t *zio); 297 298 extern void *zio_buf_alloc(size_t size); 299 extern void zio_buf_free(void *buf, size_t size); 300 extern void *zio_data_buf_alloc(size_t size); 301 extern void zio_data_buf_free(void *buf, size_t size); 302 303 /* 304 * Move an I/O to the next stage of the pipeline and execute that stage. 305 * There's no locking on io_stage because there's no legitimate way for 306 * multiple threads to be attempting to process the same I/O. 307 */ 308 extern void zio_next_stage(zio_t *zio); 309 extern void zio_next_stage_async(zio_t *zio); 310 extern void zio_wait_children_done(zio_t *zio); 311 312 /* 313 * Delegate I/O to a child vdev. 314 */ 315 extern zio_t *zio_vdev_child_io(zio_t *zio, blkptr_t *bp, vdev_t *vd, 316 uint64_t offset, void *data, uint64_t size, int type, int priority, 317 int flags, zio_done_func_t *done, void *private); 318 319 extern void zio_vdev_io_bypass(zio_t *zio); 320 extern void zio_vdev_io_reissue(zio_t *zio); 321 extern void zio_vdev_io_redone(zio_t *zio); 322 323 extern void zio_checksum_verified(zio_t *zio); 324 extern void zio_set_gang_verifier(zio_t *zio, zio_cksum_t *zcp); 325 326 extern uint8_t zio_checksum_select(uint8_t child, uint8_t parent); 327 extern uint8_t zio_compress_select(uint8_t child, uint8_t parent); 328 329 boolean_t zio_should_retry(zio_t *zio); 330 331 /* 332 * Initial setup and teardown. 333 */ 334 extern void zio_init(void); 335 extern void zio_fini(void); 336 337 /* 338 * Fault injection 339 */ 340 struct zinject_record; 341 extern uint32_t zio_injection_enabled; 342 extern int zio_inject_fault(char *name, int flags, int *id, 343 struct zinject_record *record); 344 extern int zio_inject_list_next(int *id, char *name, size_t buflen, 345 struct zinject_record *record); 346 extern int zio_clear_fault(int id); 347 extern int zio_handle_fault_injection(zio_t *zio, int error); 348 extern int zio_handle_device_injection(vdev_t *vd, int error); 349 350 #ifdef __cplusplus 351 } 352 #endif 353 354 #endif /* _ZIO_H */ 355