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 2008 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _SYS_ZAP_LEAF_H 27 #define _SYS_ZAP_LEAF_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #ifdef __cplusplus 32 extern "C" { 33 #endif 34 35 struct zap; 36 37 #define ZAP_LEAF_MAGIC 0x2AB1EAF 38 39 /* chunk size = 24 bytes */ 40 #define ZAP_LEAF_CHUNKSIZE 24 41 42 /* 43 * The amount of space available for chunks is: 44 * block size (1<<l->l_bs) - hash entry size (2) * number of hash 45 * entries - header space (2*chunksize) 46 */ 47 #define ZAP_LEAF_NUMCHUNKS(l) \ 48 (((1<<(l)->l_bs) - 2*ZAP_LEAF_HASH_NUMENTRIES(l)) / \ 49 ZAP_LEAF_CHUNKSIZE - 2) 50 51 /* 52 * The amount of space within the chunk available for the array is: 53 * chunk size - space for type (1) - space for next pointer (2) 54 */ 55 #define ZAP_LEAF_ARRAY_BYTES (ZAP_LEAF_CHUNKSIZE - 3) 56 57 #define ZAP_LEAF_ARRAY_NCHUNKS(bytes) \ 58 (((bytes)+ZAP_LEAF_ARRAY_BYTES-1)/ZAP_LEAF_ARRAY_BYTES) 59 60 /* 61 * Low water mark: when there are only this many chunks free, start 62 * growing the ptrtbl. Ideally, this should be larger than a 63 * "reasonably-sized" entry. 20 chunks is more than enough for the 64 * largest directory entry (MAXNAMELEN (256) byte name, 8-byte value), 65 * while still being only around 3% for 16k blocks. 66 */ 67 #define ZAP_LEAF_LOW_WATER (20) 68 69 /* 70 * The leaf hash table has block size / 2^5 (32) number of entries, 71 * which should be more than enough for the maximum number of entries, 72 * which is less than block size / CHUNKSIZE (24) / minimum number of 73 * chunks per entry (3). 74 */ 75 #define ZAP_LEAF_HASH_SHIFT(l) ((l)->l_bs - 5) 76 #define ZAP_LEAF_HASH_NUMENTRIES(l) (1 << ZAP_LEAF_HASH_SHIFT(l)) 77 78 /* 79 * The chunks start immediately after the hash table. The end of the 80 * hash table is at l_hash + HASH_NUMENTRIES, which we simply cast to a 81 * chunk_t. 82 */ 83 #define ZAP_LEAF_CHUNK(l, idx) \ 84 ((zap_leaf_chunk_t *) \ 85 ((l)->l_phys->l_hash + ZAP_LEAF_HASH_NUMENTRIES(l)))[idx] 86 #define ZAP_LEAF_ENTRY(l, idx) (&ZAP_LEAF_CHUNK(l, idx).l_entry) 87 88 typedef enum zap_chunk_type { 89 ZAP_CHUNK_FREE = 253, 90 ZAP_CHUNK_ENTRY = 252, 91 ZAP_CHUNK_ARRAY = 251, 92 ZAP_CHUNK_TYPE_MAX = 250 93 } zap_chunk_type_t; 94 95 #define ZLF_ENTRIES_CDSORTED (1<<0) 96 97 /* 98 * TAKE NOTE: 99 * If zap_leaf_phys_t is modified, zap_leaf_byteswap() must be modified. 100 */ 101 typedef struct zap_leaf_phys { 102 struct zap_leaf_header { 103 uint64_t lh_block_type; /* ZBT_LEAF */ 104 uint64_t lh_pad1; 105 uint64_t lh_prefix; /* hash prefix of this leaf */ 106 uint32_t lh_magic; /* ZAP_LEAF_MAGIC */ 107 uint16_t lh_nfree; /* number free chunks */ 108 uint16_t lh_nentries; /* number of entries */ 109 uint16_t lh_prefix_len; /* num bits used to id this */ 110 111 /* above is accessable to zap, below is zap_leaf private */ 112 113 uint16_t lh_freelist; /* chunk head of free list */ 114 uint8_t lh_flags; /* ZLF_* flags */ 115 uint8_t lh_pad2[11]; 116 } l_hdr; /* 2 24-byte chunks */ 117 118 /* 119 * The header is followed by a hash table with 120 * ZAP_LEAF_HASH_NUMENTRIES(zap) entries. The hash table is 121 * followed by an array of ZAP_LEAF_NUMCHUNKS(zap) 122 * zap_leaf_chunk structures. These structures are accessed 123 * with the ZAP_LEAF_CHUNK() macro. 124 */ 125 126 uint16_t l_hash[1]; 127 } zap_leaf_phys_t; 128 129 typedef union zap_leaf_chunk { 130 struct zap_leaf_entry { 131 uint8_t le_type; /* always ZAP_CHUNK_ENTRY */ 132 uint8_t le_int_size; /* size of ints */ 133 uint16_t le_next; /* next entry in hash chain */ 134 uint16_t le_name_chunk; /* first chunk of the name */ 135 uint16_t le_name_length; /* bytes in name, incl null */ 136 uint16_t le_value_chunk; /* first chunk of the value */ 137 uint16_t le_value_length; /* value length in ints */ 138 uint32_t le_cd; /* collision differentiator */ 139 uint64_t le_hash; /* hash value of the name */ 140 } l_entry; 141 struct zap_leaf_array { 142 uint8_t la_type; /* always ZAP_CHUNK_ARRAY */ 143 uint8_t la_array[ZAP_LEAF_ARRAY_BYTES]; 144 uint16_t la_next; /* next blk or CHAIN_END */ 145 } l_array; 146 struct zap_leaf_free { 147 uint8_t lf_type; /* always ZAP_CHUNK_FREE */ 148 uint8_t lf_pad[ZAP_LEAF_ARRAY_BYTES]; 149 uint16_t lf_next; /* next in free list, or CHAIN_END */ 150 } l_free; 151 } zap_leaf_chunk_t; 152 153 typedef struct zap_leaf { 154 krwlock_t l_rwlock; 155 uint64_t l_blkid; /* 1<<ZAP_BLOCK_SHIFT byte block off */ 156 int l_bs; /* block size shift */ 157 dmu_buf_t *l_dbuf; 158 zap_leaf_phys_t *l_phys; 159 } zap_leaf_t; 160 161 162 typedef struct zap_entry_handle { 163 /* below is set by zap_leaf.c and is public to zap.c */ 164 uint64_t zeh_num_integers; 165 uint64_t zeh_hash; 166 uint32_t zeh_cd; 167 uint8_t zeh_integer_size; 168 169 /* below is private to zap_leaf.c */ 170 uint16_t zeh_fakechunk; 171 uint16_t *zeh_chunkp; 172 zap_leaf_t *zeh_leaf; 173 } zap_entry_handle_t; 174 175 /* 176 * Return a handle to the named entry, or ENOENT if not found. The hash 177 * value must equal zap_hash(name). 178 */ 179 extern int zap_leaf_lookup(zap_leaf_t *l, 180 zap_name_t *zn, zap_entry_handle_t *zeh); 181 182 /* 183 * Return a handle to the entry with this hash+cd, or the entry with the 184 * next closest hash+cd. 185 */ 186 extern int zap_leaf_lookup_closest(zap_leaf_t *l, 187 uint64_t hash, uint32_t cd, zap_entry_handle_t *zeh); 188 189 /* 190 * Read the first num_integers in the attribute. Integer size 191 * conversion will be done without sign extension. Return EINVAL if 192 * integer_size is too small. Return EOVERFLOW if there are more than 193 * num_integers in the attribute. 194 */ 195 extern int zap_entry_read(const zap_entry_handle_t *zeh, 196 uint8_t integer_size, uint64_t num_integers, void *buf); 197 198 extern int zap_entry_read_name(const zap_entry_handle_t *zeh, 199 uint16_t buflen, char *buf); 200 201 /* 202 * Replace the value of an existing entry. 203 * 204 * zap_entry_update may fail if it runs out of space (ENOSPC). 205 */ 206 extern int zap_entry_update(zap_entry_handle_t *zeh, 207 uint8_t integer_size, uint64_t num_integers, const void *buf); 208 209 /* 210 * Remove an entry. 211 */ 212 extern void zap_entry_remove(zap_entry_handle_t *zeh); 213 214 /* 215 * Create an entry. An equal entry must not exist, and this entry must 216 * belong in this leaf (according to its hash value). Fills in the 217 * entry handle on success. Returns 0 on success or ENOSPC on failure. 218 */ 219 extern int zap_entry_create(zap_leaf_t *l, 220 const char *name, uint64_t h, uint32_t cd, 221 uint8_t integer_size, uint64_t num_integers, const void *buf, 222 zap_entry_handle_t *zeh); 223 224 /* 225 * Return true if there are additional entries with the same normalized 226 * form. 227 */ 228 extern boolean_t zap_entry_normalization_conflict(zap_entry_handle_t *zeh, 229 zap_name_t *zn, const char *name, zap_t *zap); 230 231 /* 232 * Other stuff. 233 */ 234 235 extern void zap_leaf_init(zap_leaf_t *l, boolean_t sort); 236 extern void zap_leaf_byteswap(zap_leaf_phys_t *buf, int len); 237 extern void zap_leaf_split(zap_leaf_t *l, zap_leaf_t *nl, boolean_t sort); 238 extern void zap_leaf_stats(zap_t *zap, zap_leaf_t *l, zap_stats_t *zs); 239 240 #ifdef __cplusplus 241 } 242 #endif 243 244 #endif /* _SYS_ZAP_LEAF_H */ 245