include/sys/zap_leaf.h

   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 https://opensource.org/licenses/CDDL-1.0.
  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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  23  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
  24  */
  25
  26 #ifndef _SYS_ZAP_LEAF_H
  27 #define _SYS_ZAP_LEAF_H
  28
  29 #include <sys/zap.h>
  30
  31 #ifdef  __cplusplus
  32 extern "C" {
  33 #endif
  34
  35 struct zap;
  36 struct zap_name;
  37 struct zap_stats;
  38
  39 #define ZAP_LEAF_MAGIC 0x2AB1EAF
  40
  41 /* chunk size = 24 bytes */
  42 #define ZAP_LEAF_CHUNKSIZE 24
  43
  44 /*
  45  * The amount of space available for chunks is:
  46  * block size (1<<l->l_bs) - hash entry size (2) * number of hash
  47  * entries - header space (2*chunksize)
  48  */
  49 #define ZAP_LEAF_NUMCHUNKS_BS(bs) \
  50         (((1U << (bs)) - 2 * ZAP_LEAF_HASH_NUMENTRIES_BS(bs)) / \
  51         ZAP_LEAF_CHUNKSIZE - 2)
  52
  53 #define ZAP_LEAF_NUMCHUNKS(l) (ZAP_LEAF_NUMCHUNKS_BS(((l)->l_bs)))
  54
  55 #define ZAP_LEAF_NUMCHUNKS_DEF \
  56         (ZAP_LEAF_NUMCHUNKS_BS(fzap_default_block_shift))
  57
  58 /*
  59  * The amount of space within the chunk available for the array is:
  60  * chunk size - space for type (1) - space for next pointer (2)
  61  */
  62 #define ZAP_LEAF_ARRAY_BYTES (ZAP_LEAF_CHUNKSIZE - 3)
  63
  64 #define ZAP_LEAF_ARRAY_NCHUNKS(bytes) \
  65         (((bytes)+ZAP_LEAF_ARRAY_BYTES-1)/ZAP_LEAF_ARRAY_BYTES)
  66
  67 /*
  68  * Low water mark:  when there are only this many chunks free, start
  69  * growing the ptrtbl.  Ideally, this should be larger than a
  70  * "reasonably-sized" entry.  20 chunks is more than enough for the
  71  * largest directory entry (MAXNAMELEN (256) byte name, 8-byte value),
  72  * while still being only around 3% for 16k blocks.
  73  */
  74 #define ZAP_LEAF_LOW_WATER (20)
  75
  76 /*
  77  * The leaf hash table has block size / 2^5 (32) number of entries,
  78  * which should be more than enough for the maximum number of entries,
  79  * which is less than block size / CHUNKSIZE (24) / minimum number of
  80  * chunks per entry (3).
  81  */
  82 #define ZAP_LEAF_HASH_SHIFT_BS(bs) ((bs) - 5)
  83 #define ZAP_LEAF_HASH_NUMENTRIES_BS(bs) (1U << ZAP_LEAF_HASH_SHIFT_BS(bs))
  84 #define ZAP_LEAF_HASH_SHIFT(l) (ZAP_LEAF_HASH_SHIFT_BS(((l)->l_bs)))
  85 #define ZAP_LEAF_HASH_NUMENTRIES(l) (ZAP_LEAF_HASH_NUMENTRIES_BS(((l)->l_bs)))
  86
  87 /*
  88  * The chunks start immediately after the hash table.  The end of the
  89  * hash table is at l_hash + HASH_NUMENTRIES, which we simply cast to a
  90  * chunk_t.
  91  */
  92 #define ZAP_LEAF_CHUNK(l, idx) \
  93         ((zap_leaf_chunk_t *) \
  94         (zap_leaf_phys(l)->l_hash + ZAP_LEAF_HASH_NUMENTRIES(l)))[idx]
  95 #define ZAP_LEAF_ENTRY(l, idx) (&ZAP_LEAF_CHUNK(l, idx).l_entry)
  96
  97 typedef enum zap_chunk_type {
  98         ZAP_CHUNK_FREE = 253,
  99         ZAP_CHUNK_ENTRY = 252,
 100         ZAP_CHUNK_ARRAY = 251,
 101         ZAP_CHUNK_TYPE_MAX = 250
 102 } zap_chunk_type_t;
 103
 104 #define ZLF_ENTRIES_CDSORTED (1<<0)
 105
 106 /*
 107  * TAKE NOTE:
 108  * If zap_leaf_phys_t is modified, zap_leaf_byteswap() must be modified.
 109  */
 110 typedef struct zap_leaf_phys {
 111         struct zap_leaf_header {
 112                 /* Public to ZAP */
 113                 uint64_t lh_block_type;         /* ZBT_LEAF */
 114                 uint64_t lh_pad1;
 115                 uint64_t lh_prefix;             /* hash prefix of this leaf */
 116                 uint32_t lh_magic;              /* ZAP_LEAF_MAGIC */
 117                 uint16_t lh_nfree;              /* number free chunks */
 118                 uint16_t lh_nentries;           /* number of entries */
 119                 uint16_t lh_prefix_len;         /* num bits used to id this */
 120
 121                 /* Private to zap_leaf */
 122                 uint16_t lh_freelist;           /* chunk head of free list */
 123                 uint8_t lh_flags;               /* ZLF_* flags */
 124                 uint8_t lh_pad2[11];
 125         } l_hdr; /* 2 24-byte chunks */
 126
 127         /*
 128          * The header is followed by a hash table with
 129          * ZAP_LEAF_HASH_NUMENTRIES(zap) entries.  The hash table is
 130          * followed by an array of ZAP_LEAF_NUMCHUNKS(zap)
 131          * zap_leaf_chunk structures.  These structures are accessed
 132          * with the ZAP_LEAF_CHUNK() macro.
 133          */
 134
 135         uint16_t l_hash[];
 136 } zap_leaf_phys_t;
 137
 138 typedef union zap_leaf_chunk {
 139         struct zap_leaf_entry {
 140                 uint8_t le_type;                /* always ZAP_CHUNK_ENTRY */
 141                 uint8_t le_value_intlen;        /* size of value's ints */
 142                 uint16_t le_next;               /* next entry in hash chain */
 143                 uint16_t le_name_chunk;         /* first chunk of the name */
 144                 uint16_t le_name_numints;       /* ints in name (incl null) */
 145                 uint16_t le_value_chunk;        /* first chunk of the value */
 146                 uint16_t le_value_numints;      /* value length in ints */
 147                 uint32_t le_cd;                 /* collision differentiator */
 148                 uint64_t le_hash;               /* hash value of the name */
 149         } l_entry;
 150         struct zap_leaf_array {
 151                 uint8_t la_type;                /* always ZAP_CHUNK_ARRAY */
 152                 uint8_t la_array[ZAP_LEAF_ARRAY_BYTES];
 153                 uint16_t la_next;               /* next blk or CHAIN_END */
 154         } l_array;
 155         struct zap_leaf_free {
 156                 uint8_t lf_type;                /* always ZAP_CHUNK_FREE */
 157                 uint8_t lf_pad[ZAP_LEAF_ARRAY_BYTES];
 158                 uint16_t lf_next;       /* next in free list, or CHAIN_END */
 159         } l_free;
 160 } zap_leaf_chunk_t;
 161
 162 typedef struct zap_leaf {
 163         dmu_buf_user_t l_dbu;
 164         krwlock_t l_rwlock;
 165         uint64_t l_blkid;               /* 1<<ZAP_BLOCK_SHIFT byte block off */
 166         uint_t l_bs;                    /* block size shift */
 167         dmu_buf_t *l_dbuf;
 168 } zap_leaf_t;
 169
 170 static inline zap_leaf_phys_t *
 171 zap_leaf_phys(zap_leaf_t *l)
 172 {
 173         return (l->l_dbuf->db_data);
 174 }
 175
 176 typedef struct zap_entry_handle {
 177         /* Set by zap_leaf and public to ZAP */
 178         uint64_t zeh_num_integers;
 179         uint64_t zeh_hash;
 180         uint32_t zeh_cd;
 181         uint8_t zeh_integer_size;
 182
 183         /* Private to zap_leaf */
 184         uint16_t zeh_fakechunk;
 185         uint16_t *zeh_chunkp;
 186         zap_leaf_t *zeh_leaf;
 187 } zap_entry_handle_t;
 188
 189 /*
 190  * Return a handle to the named entry, or ENOENT if not found.  The hash
 191  * value must equal zap_hash(name).
 192  */
 193 extern int zap_leaf_lookup(zap_leaf_t *l,
 194     struct zap_name *zn, zap_entry_handle_t *zeh);
 195
 196 /*
 197  * Return a handle to the entry with this hash+cd, or the entry with the
 198  * next closest hash+cd.
 199  */
 200 extern int zap_leaf_lookup_closest(zap_leaf_t *l,
 201     uint64_t hash, uint32_t cd, zap_entry_handle_t *zeh);
 202
 203 /*
 204  * Read the first num_integers in the attribute.  Integer size
 205  * conversion will be done without sign extension.  Return EINVAL if
 206  * integer_size is too small.  Return EOVERFLOW if there are more than
 207  * num_integers in the attribute.
 208  */
 209 extern int zap_entry_read(const zap_entry_handle_t *zeh,
 210     uint8_t integer_size, uint64_t num_integers, void *buf);
 211
 212 extern int zap_entry_read_name(struct zap *zap, const zap_entry_handle_t *zeh,
 213     uint16_t buflen, char *buf);
 214
 215 /*
 216  * Replace the value of an existing entry.
 217  *
 218  * May fail if it runs out of space (ENOSPC).
 219  */
 220 extern int zap_entry_update(zap_entry_handle_t *zeh,
 221     uint8_t integer_size, uint64_t num_integers, const void *buf);
 222
 223 /*
 224  * Remove an entry.
 225  */
 226 extern void zap_entry_remove(zap_entry_handle_t *zeh);
 227
 228 /*
 229  * Create an entry. An equal entry must not exist, and this entry must
 230  * belong in this leaf (according to its hash value).  Fills in the
 231  * entry handle on success.  Returns 0 on success or ENOSPC on failure.
 232  */
 233 extern int zap_entry_create(zap_leaf_t *l, struct zap_name *zn, uint32_t cd,
 234     uint8_t integer_size, uint64_t num_integers, const void *buf,
 235     zap_entry_handle_t *zeh);
 236
 237 /* Determine whether there is another entry with the same normalized form. */
 238 extern boolean_t zap_entry_normalization_conflict(zap_entry_handle_t *zeh,
 239     struct zap_name *zn, const char *name, struct zap *zap);
 240
 241 /*
 242  * Other stuff.
 243  */
 244
 245 extern void zap_leaf_init(zap_leaf_t *l, boolean_t sort);
 246 extern void zap_leaf_byteswap(zap_leaf_phys_t *buf, size_t len);
 247 extern void zap_leaf_split(zap_leaf_t *l, zap_leaf_t *nl, boolean_t sort);
 248 extern void zap_leaf_stats(struct zap *zap, zap_leaf_t *l,
 249     struct zap_stats *zs);
 250
 251 #ifdef  __cplusplus
 252 }
 253 #endif
 254
 255 #endif /* _SYS_ZAP_LEAF_H */