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