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 2006 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _SYS_ZAP_H 27 #define _SYS_ZAP_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 /* 32 * ZAP - ZFS Attribute Processor 33 * 34 * The ZAP is a module which sits on top of the DMU (Data Managemnt 35 * Unit) and implements a higher-level storage primitive using DMU 36 * objects. Its primary consumer is the ZPL (ZFS Posix Layer). 37 * 38 * A "zapobj" is a DMU object which the ZAP uses to stores attributes. 39 * Users should use only zap routines to access a zapobj - they should 40 * not access the DMU object directly using DMU routines. 41 * 42 * The attributes stored in a zapobj are name-value pairs. The name is 43 * a zero-terminated string of up to ZAP_MAXNAMELEN bytes (including 44 * terminating NULL). The value is an array of integers, which may be 45 * 1, 2, 4, or 8 bytes long. The total space used by the array (number 46 * of integers * integer length) can be up to ZAP_MAXVALUELEN bytes. 47 * Note that an 8-byte integer value can be used to store the location 48 * (object number) of another dmu object (which may be itself a zapobj). 49 * Note that you can use a zero-length attribute to store a single bit 50 * of information - the attribute is present or not. 51 * 52 * The ZAP routines are thread-safe. However, you must observe the 53 * DMU's restriction that a transaction may not be operated on 54 * concurrently. 55 * 56 * Any of the routines that return an int may return an I/O error (EIO 57 * or ECHECKSUM). 58 * 59 * 60 * Implementation / Performance Notes: 61 * 62 * The ZAP is intended to operate most efficiently on attributes with 63 * short (49 bytes or less) names and single 8-byte values, for which 64 * the microzap will be used. The ZAP should be efficient enough so 65 * that the user does not need to cache these attributes. 66 * 67 * The ZAP's locking scheme makes its routines thread-safe. Operations 68 * on different zapobjs will be processed concurrently. Operations on 69 * the same zapobj which only read data will be processed concurrently. 70 * Operations on the same zapobj which modify data will be processed 71 * concurrently when there are many attributes in the zapobj (because 72 * the ZAP uses per-block locking - more than 128 * (number of cpus) 73 * small attributes will suffice). 74 */ 75 76 /* 77 * We're using zero-terminated byte strings (ie. ASCII or UTF-8 C 78 * strings) for the names of attributes, rather than a byte string 79 * bounded by an explicit length. If some day we want to support names 80 * in character sets which have embedded zeros (eg. UTF-16, UTF-32), 81 * we'll have to add routines for using length-bounded strings. 82 */ 83 84 #include <sys/dmu.h> 85 86 #ifdef __cplusplus 87 extern "C" { 88 #endif 89 90 #define ZAP_MAXNAMELEN 256 91 #define ZAP_MAXVALUELEN 1024 92 93 /* 94 * Create a new zapobj with no attributes and return its object number. 95 */ 96 uint64_t zap_create(objset_t *ds, dmu_object_type_t ot, 97 dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx); 98 99 /* 100 * Create a new zapobj with no attributes from the given (unallocated) 101 * object number. 102 */ 103 int zap_create_claim(objset_t *ds, uint64_t obj, dmu_object_type_t ot, 104 dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx); 105 106 /* 107 * The zapobj passed in must be a valid ZAP object for all of the 108 * following routines. 109 */ 110 111 /* 112 * Destroy this zapobj and all its attributes. 113 * 114 * Frees the object number using dmu_object_free. 115 */ 116 int zap_destroy(objset_t *ds, uint64_t zapobj, dmu_tx_t *tx); 117 118 /* 119 * Manipulate attributes. 120 * 121 * 'integer_size' is in bytes, and must be 1, 2, 4, or 8. 122 */ 123 124 /* 125 * Retrieve the contents of the attribute with the given name. 126 * 127 * If the requested attribute does not exist, the call will fail and 128 * return ENOENT. 129 * 130 * If 'integer_size' is smaller than the attribute's integer size, the 131 * call will fail and return EINVAL. 132 * 133 * If 'integer_size' is equal to or larger than the attribute's integer 134 * size, the call will succeed and return 0. * When converting to a 135 * larger integer size, the integers will be treated as unsigned (ie. no 136 * sign-extension will be performed). 137 * 138 * 'num_integers' is the length (in integers) of 'buf'. 139 * 140 * If the attribute is longer than the buffer, as many integers as will 141 * fit will be transferred to 'buf'. If the entire attribute was not 142 * transferred, the call will return EOVERFLOW. 143 */ 144 int zap_lookup(objset_t *ds, uint64_t zapobj, const char *name, 145 uint64_t integer_size, uint64_t num_integers, void *buf); 146 147 /* 148 * Create an attribute with the given name and value. 149 * 150 * If an attribute with the given name already exists, the call will 151 * fail and return EEXIST. 152 */ 153 int zap_add(objset_t *ds, uint64_t zapobj, const char *name, 154 int integer_size, uint64_t num_integers, 155 const void *val, dmu_tx_t *tx); 156 157 /* 158 * Set the attribute with the given name to the given value. If an 159 * attribute with the given name does not exist, it will be created. If 160 * an attribute with the given name already exists, the previous value 161 * will be overwritten. The integer_size may be different from the 162 * existing attribute's integer size, in which case the attribute's 163 * integer size will be updated to the new value. 164 */ 165 int zap_update(objset_t *ds, uint64_t zapobj, const char *name, 166 int integer_size, uint64_t num_integers, const void *val, dmu_tx_t *tx); 167 168 /* 169 * Get the length (in integers) and the integer size of the specified 170 * attribute. 171 * 172 * If the requested attribute does not exist, the call will fail and 173 * return ENOENT. 174 */ 175 int zap_length(objset_t *ds, uint64_t zapobj, const char *name, 176 uint64_t *integer_size, uint64_t *num_integers); 177 178 /* 179 * Remove the specified attribute. 180 * 181 * If the specified attribute does not exist, the call will fail and 182 * return ENOENT. 183 */ 184 int zap_remove(objset_t *ds, uint64_t zapobj, const char *name, dmu_tx_t *tx); 185 186 /* 187 * Returns (in *count) the number of attributes in the specified zap 188 * object. 189 */ 190 int zap_count(objset_t *ds, uint64_t zapobj, uint64_t *count); 191 192 193 /* 194 * Returns (in name) the name of the entry whose value 195 * (za_first_integer) is value, or ENOENT if not found. The string 196 * pointed to by name must be at least 256 bytes long. 197 */ 198 int zap_value_search(objset_t *os, uint64_t zapobj, uint64_t value, char *name); 199 200 struct zap; 201 struct zap_leaf; 202 typedef struct zap_cursor { 203 /* This structure is opaque! */ 204 objset_t *zc_objset; 205 struct zap *zc_zap; 206 struct zap_leaf *zc_leaf; 207 uint64_t zc_zapobj; 208 uint64_t zc_hash; 209 uint32_t zc_cd; 210 } zap_cursor_t; 211 212 typedef struct { 213 int za_integer_length; 214 uint64_t za_num_integers; 215 uint64_t za_first_integer; /* no sign extension for <8byte ints */ 216 char za_name[MAXNAMELEN]; 217 } zap_attribute_t; 218 219 /* 220 * The interface for listing all the attributes of a zapobj can be 221 * thought of as cursor moving down a list of the attributes one by 222 * one. The cookie returned by the zap_cursor_serialize routine is 223 * persistent across system calls (and across reboot, even). 224 */ 225 226 /* 227 * Initialize a zap cursor, pointing to the "first" attribute of the 228 * zapobj. You must _fini the cursor when you are done with it. 229 */ 230 void zap_cursor_init(zap_cursor_t *zc, objset_t *ds, uint64_t zapobj); 231 void zap_cursor_fini(zap_cursor_t *zc); 232 233 /* 234 * Get the attribute currently pointed to by the cursor. Returns 235 * ENOENT if at the end of the attributes. 236 */ 237 int zap_cursor_retrieve(zap_cursor_t *zc, zap_attribute_t *za); 238 239 /* 240 * Advance the cursor to the next attribute. 241 */ 242 void zap_cursor_advance(zap_cursor_t *zc); 243 244 /* 245 * Get a persistent cookie pointing to the current position of the zap 246 * cursor. The low 4 bits in the cookie are always zero, and thus can 247 * be used as to differentiate a serialized cookie from a different type 248 * of value. The cookie will be less than 2^32 as long as there are 249 * fewer than 2^22 (4.2 million) entries in the zap object. 250 */ 251 uint64_t zap_cursor_serialize(zap_cursor_t *zc); 252 253 /* 254 * Initialize a zap cursor pointing to the position recorded by 255 * zap_cursor_serialize (in the "serialized" argument). You can also 256 * use a "serialized" argument of 0 to start at the beginning of the 257 * zapobj (ie. zap_cursor_init_serialized(..., 0) is equivalent to 258 * zap_cursor_init(...).) 259 */ 260 void zap_cursor_init_serialized(zap_cursor_t *zc, objset_t *ds, 261 uint64_t zapobj, uint64_t serialized); 262 263 264 #define ZAP_HISTOGRAM_SIZE 10 265 266 typedef struct zap_stats { 267 /* 268 * Size of the pointer table (in number of entries). 269 * This is always a power of 2, or zero if it's a microzap. 270 * In general, it should be considerably greater than zs_num_leafs. 271 */ 272 uint64_t zs_ptrtbl_len; 273 274 uint64_t zs_blocksize; /* size of zap blocks */ 275 276 /* 277 * The number of blocks used. Note that some blocks may be 278 * wasted because old ptrtbl's and large name/value blocks are 279 * not reused. (Although their space is reclaimed, we don't 280 * reuse those offsets in the object.) 281 */ 282 uint64_t zs_num_blocks; 283 284 /* 285 * Pointer table values from zap_ptrtbl in the zap_phys_t 286 */ 287 uint64_t zs_ptrtbl_nextblk; /* next (larger) copy start block */ 288 uint64_t zs_ptrtbl_blks_copied; /* number source blocks copied */ 289 uint64_t zs_ptrtbl_zt_blk; /* starting block number */ 290 uint64_t zs_ptrtbl_zt_numblks; /* number of blocks */ 291 uint64_t zs_ptrtbl_zt_shift; /* bits to index it */ 292 293 /* 294 * Values of the other members of the zap_phys_t 295 */ 296 uint64_t zs_block_type; /* ZBT_HEADER */ 297 uint64_t zs_magic; /* ZAP_MAGIC */ 298 uint64_t zs_num_leafs; /* The number of leaf blocks */ 299 uint64_t zs_num_entries; /* The number of zap entries */ 300 uint64_t zs_salt; /* salt to stir into hash function */ 301 302 /* 303 * Histograms. For all histograms, the last index 304 * (ZAP_HISTOGRAM_SIZE-1) includes any values which are greater 305 * than what can be represented. For example 306 * zs_leafs_with_n5_entries[ZAP_HISTOGRAM_SIZE-1] is the number 307 * of leafs with more than 45 entries. 308 */ 309 310 /* 311 * zs_leafs_with_n_pointers[n] is the number of leafs with 312 * 2^n pointers to it. 313 */ 314 uint64_t zs_leafs_with_2n_pointers[ZAP_HISTOGRAM_SIZE]; 315 316 /* 317 * zs_leafs_with_n_entries[n] is the number of leafs with 318 * [n*5, (n+1)*5) entries. In the current implementation, there 319 * can be at most 55 entries in any block, but there may be 320 * fewer if the name or value is large, or the block is not 321 * completely full. 322 */ 323 uint64_t zs_blocks_with_n5_entries[ZAP_HISTOGRAM_SIZE]; 324 325 /* 326 * zs_leafs_n_tenths_full[n] is the number of leafs whose 327 * fullness is in the range [n/10, (n+1)/10). 328 */ 329 uint64_t zs_blocks_n_tenths_full[ZAP_HISTOGRAM_SIZE]; 330 331 /* 332 * zs_entries_using_n_chunks[n] is the number of entries which 333 * consume n 24-byte chunks. (Note, large names/values only use 334 * one chunk, but contribute to zs_num_blocks_large.) 335 */ 336 uint64_t zs_entries_using_n_chunks[ZAP_HISTOGRAM_SIZE]; 337 338 /* 339 * zs_buckets_with_n_entries[n] is the number of buckets (each 340 * leaf has 64 buckets) with n entries. 341 * zs_buckets_with_n_entries[1] should be very close to 342 * zs_num_entries. 343 */ 344 uint64_t zs_buckets_with_n_entries[ZAP_HISTOGRAM_SIZE]; 345 } zap_stats_t; 346 347 /* 348 * Get statistics about a ZAP object. Note: you need to be aware of the 349 * internal implementation of the ZAP to correctly interpret some of the 350 * statistics. This interface shouldn't be relied on unless you really 351 * know what you're doing. 352 */ 353 int zap_get_stats(objset_t *ds, uint64_t zapobj, zap_stats_t *zs); 354 355 #ifdef __cplusplus 356 } 357 #endif 358 359 #endif /* _SYS_ZAP_H */ 360