src/include/access/hash.h

   1 /*-------------------------------------------------------------------------
   2  *
   3  * hash.h
   4  *        header file for postgres hash access method implementation
   5  *
   6  *
   7  * Portions Copyright (c) 1996-2009, PostgreSQL Global Development Group
   8  * Portions Copyright (c) 1994, Regents of the University of California
   9  *
  10  * $PostgreSQL$
  11  *
  12  * NOTES
  13  *              modeled after Margo Seltzer's hash implementation for unix.
  14  *
  15  *-------------------------------------------------------------------------
  16  */
  17 #ifndef HASH_H
  18 #define HASH_H
  19
  20 #include "access/genam.h"
  21 #include "access/itup.h"
  22 #include "access/sdir.h"
  23 #include "access/xlog.h"
  24 #include "fmgr.h"
  25 #include "storage/lock.h"
  26 #include "utils/relcache.h"
  27
  28 /*
  29  * Mapping from hash bucket number to physical block number of bucket's
  30  * starting page.  Beware of multiple evaluations of argument!
  31  */
  32 typedef uint32 Bucket;
  33
  34 #define BUCKET_TO_BLKNO(metap,B) \
  35                 ((BlockNumber) ((B) + ((B) ? (metap)->hashm_spares[_hash_log2((B)+1)-1] : 0)) + 1)
  36
  37 /*
  38  * Special space for hash index pages.
  39  *
  40  * hasho_flag tells us which type of page we're looking at.  For
  41  * example, knowing overflow pages from bucket pages is necessary
  42  * information when you're deleting tuples from a page. If all the
  43  * tuples are deleted from an overflow page, the overflow is made
  44  * available to other buckets by calling _hash_freeovflpage(). If all
  45  * the tuples are deleted from a bucket page, no additional action is
  46  * necessary.
  47  */
  48 #define LH_UNUSED_PAGE                  (0)
  49 #define LH_OVERFLOW_PAGE                (1 << 0)
  50 #define LH_BUCKET_PAGE                  (1 << 1)
  51 #define LH_BITMAP_PAGE                  (1 << 2)
  52 #define LH_META_PAGE                    (1 << 3)
  53
  54 typedef struct HashPageOpaqueData
  55 {
  56         BlockNumber hasho_prevblkno;    /* previous ovfl (or bucket) blkno */
  57         BlockNumber hasho_nextblkno;    /* next ovfl blkno */
  58         Bucket          hasho_bucket;   /* bucket number this pg belongs to */
  59         uint16          hasho_flag;             /* page type code, see above */
  60         uint16          hasho_page_id;  /* for identification of hash indexes */
  61 } HashPageOpaqueData;
  62
  63 typedef HashPageOpaqueData *HashPageOpaque;
  64
  65 /*
  66  * The page ID is for the convenience of pg_filedump and similar utilities,
  67  * which otherwise would have a hard time telling pages of different index
  68  * types apart.  It should be the last 2 bytes on the page.  This is more or
  69  * less "free" due to alignment considerations.
  70  */
  71 #define HASHO_PAGE_ID           0xFF80
  72
  73 /*
  74  *      HashScanOpaqueData is private state for a hash index scan.
  75  */
  76 typedef struct HashScanOpaqueData
  77 {
  78         /* Hash value of the scan key, ie, the hash key we seek */
  79         uint32          hashso_sk_hash;
  80
  81         /*
  82          * By definition, a hash scan should be examining only one bucket. We
  83          * record the bucket number here as soon as it is known.
  84          */
  85         Bucket          hashso_bucket;
  86         bool            hashso_bucket_valid;
  87
  88         /*
  89          * If we have a share lock on the bucket, we record it here.  When
  90          * hashso_bucket_blkno is zero, we have no such lock.
  91          */
  92         BlockNumber hashso_bucket_blkno;
  93
  94         /*
  95          * We also want to remember which buffer we're currently examining in the
  96          * scan. We keep the buffer pinned (but not locked) across hashgettuple
  97          * calls, in order to avoid doing a ReadBuffer() for every tuple in the
  98          * index.
  99          */
 100         Buffer          hashso_curbuf;
 101
 102         /* Current position of the scan */
 103         ItemPointerData hashso_curpos;
 104 } HashScanOpaqueData;
 105
 106 typedef HashScanOpaqueData *HashScanOpaque;
 107
 108 /*
 109  * Definitions for metapage.
 110  */
 111
 112 #define HASH_METAPAGE   0               /* metapage is always block 0 */
 113
 114 #define HASH_MAGIC              0x6440640
 115 #define HASH_VERSION    2               /* 2 signifies only hash key value is stored */
 116
 117 /*
 118  * Spares[] holds the number of overflow pages currently allocated at or
 119  * before a certain splitpoint. For example, if spares[3] = 7 then there are
 120  * 7 ovflpages before splitpoint 3 (compare BUCKET_TO_BLKNO macro).  The
 121  * value in spares[ovflpoint] increases as overflow pages are added at the
 122  * end of the index.  Once ovflpoint increases (ie, we have actually allocated
 123  * the bucket pages belonging to that splitpoint) the number of spares at the
 124  * prior splitpoint cannot change anymore.
 125  *
 126  * ovflpages that have been recycled for reuse can be found by looking at
 127  * bitmaps that are stored within ovflpages dedicated for the purpose.
 128  * The blknos of these bitmap pages are kept in bitmaps[]; nmaps is the
 129  * number of currently existing bitmaps.
 130  *
 131  * The limitation on the size of spares[] comes from the fact that there's
 132  * no point in having more than 2^32 buckets with only uint32 hashcodes.
 133  * There is no particular upper limit on the size of mapp[], other than
 134  * needing to fit into the metapage.  (With 8K block size, 128 bitmaps
 135  * limit us to 64 Gb of overflow space...)
 136  */
 137 #define HASH_MAX_SPLITPOINTS            32
 138 #define HASH_MAX_BITMAPS                        128
 139
 140 typedef struct HashMetaPageData
 141 {
 142         uint32          hashm_magic;    /* magic no. for hash tables */
 143         uint32          hashm_version;  /* version ID */
 144         double          hashm_ntuples;  /* number of tuples stored in the table */
 145         uint16          hashm_ffactor;  /* target fill factor (tuples/bucket) */
 146         uint16          hashm_bsize;    /* index page size (bytes) */
 147         uint16          hashm_bmsize;   /* bitmap array size (bytes) - must be a power
 148                                                                  * of 2 */
 149         uint16          hashm_bmshift;  /* log2(bitmap array size in BITS) */
 150         uint32          hashm_maxbucket;        /* ID of maximum bucket in use */
 151         uint32          hashm_highmask; /* mask to modulo into entire table */
 152         uint32          hashm_lowmask;  /* mask to modulo into lower half of table */
 153         uint32          hashm_ovflpoint;/* splitpoint from which ovflpgs being
 154                                                                  * allocated */
 155         uint32          hashm_firstfree;        /* lowest-number free ovflpage (bit#) */
 156         uint32          hashm_nmaps;    /* number of bitmap pages */
 157         RegProcedure hashm_procid;      /* hash procedure id from pg_proc */
 158         uint32          hashm_spares[HASH_MAX_SPLITPOINTS];             /* spare pages before
 159                                                                                                                  * each splitpoint */
 160         BlockNumber hashm_mapp[HASH_MAX_BITMAPS];       /* blknos of ovfl bitmaps */
 161 } HashMetaPageData;
 162
 163 typedef HashMetaPageData *HashMetaPage;
 164
 165 /*
 166  * Maximum size of a hash index item (it's okay to have only one per page)
 167  */
 168 #define HashMaxItemSize(page) \
 169         MAXALIGN_DOWN(PageGetPageSize(page) - \
 170                                   SizeOfPageHeaderData - \
 171                                   sizeof(ItemIdData) - \
 172                                   MAXALIGN(sizeof(HashPageOpaqueData)))
 173
 174 #define HASH_MIN_FILLFACTOR                     10
 175 #define HASH_DEFAULT_FILLFACTOR         75
 176
 177 /*
 178  * Constants
 179  */
 180 #define BYTE_TO_BIT                             3               /* 2^3 bits/byte */
 181 #define ALL_SET                                 ((uint32) ~0)
 182
 183 /*
 184  * Bitmap pages do not contain tuples.  They do contain the standard
 185  * page headers and trailers; however, everything in between is a
 186  * giant bit array.  The number of bits that fit on a page obviously
 187  * depends on the page size and the header/trailer overhead.  We require
 188  * the number of bits per page to be a power of 2.
 189  */
 190 #define BMPGSZ_BYTE(metap)              ((metap)->hashm_bmsize)
 191 #define BMPGSZ_BIT(metap)               ((metap)->hashm_bmsize << BYTE_TO_BIT)
 192 #define BMPG_SHIFT(metap)               ((metap)->hashm_bmshift)
 193 #define BMPG_MASK(metap)                (BMPGSZ_BIT(metap) - 1)
 194
 195 #define HashPageGetBitmap(page) \
 196         ((uint32 *) PageGetContents(page))
 197
 198 #define HashGetMaxBitmapSize(page) \
 199         (PageGetPageSize((Page) page) - \
 200          (MAXALIGN(SizeOfPageHeaderData) + MAXALIGN(sizeof(HashPageOpaqueData))))
 201
 202 #define HashPageGetMeta(page) \
 203         ((HashMetaPage) PageGetContents(page))
 204
 205 /*
 206  * The number of bits in an ovflpage bitmap word.
 207  */
 208 #define BITS_PER_MAP    32              /* Number of bits in uint32 */
 209
 210 /* Given the address of the beginning of a bit map, clear/set the nth bit */
 211 #define CLRBIT(A, N)    ((A)[(N)/BITS_PER_MAP] &= ~(1<<((N)%BITS_PER_MAP)))
 212 #define SETBIT(A, N)    ((A)[(N)/BITS_PER_MAP] |= (1<<((N)%BITS_PER_MAP)))
 213 #define ISSET(A, N)             ((A)[(N)/BITS_PER_MAP] & (1<<((N)%BITS_PER_MAP)))
 214
 215 /*
 216  * page-level and high-level locking modes (see README)
 217  */
 218 #define HASH_READ               BUFFER_LOCK_SHARE
 219 #define HASH_WRITE              BUFFER_LOCK_EXCLUSIVE
 220 #define HASH_NOLOCK             (-1)
 221
 222 #define HASH_SHARE              ShareLock
 223 #define HASH_EXCLUSIVE  ExclusiveLock
 224
 225 /*
 226  *      Strategy number. There's only one valid strategy for hashing: equality.
 227  */
 228 #define HTEqualStrategyNumber                   1
 229 #define HTMaxStrategyNumber                             1
 230
 231 /*
 232  *      When a new operator class is declared, we require that the user supply
 233  *      us with an amproc procudure for hashing a key of the new type.
 234  *      Since we only have one such proc in amproc, it's number 1.
 235  */
 236 #define HASHPROC                1
 237
 238
 239 /* public routines */
 240
 241 extern Datum hashbuild(PG_FUNCTION_ARGS);
 242 extern Datum hashinsert(PG_FUNCTION_ARGS);
 243 extern Datum hashbeginscan(PG_FUNCTION_ARGS);
 244 extern Datum hashgettuple(PG_FUNCTION_ARGS);
 245 extern Datum hashgetbitmap(PG_FUNCTION_ARGS);
 246 extern Datum hashrescan(PG_FUNCTION_ARGS);
 247 extern Datum hashendscan(PG_FUNCTION_ARGS);
 248 extern Datum hashmarkpos(PG_FUNCTION_ARGS);
 249 extern Datum hashrestrpos(PG_FUNCTION_ARGS);
 250 extern Datum hashbulkdelete(PG_FUNCTION_ARGS);
 251 extern Datum hashvacuumcleanup(PG_FUNCTION_ARGS);
 252 extern Datum hashoptions(PG_FUNCTION_ARGS);
 253
 254 /*
 255  * Datatype-specific hash functions in hashfunc.c.
 256  *
 257  * These support both hash indexes and hash joins.
 258  *
 259  * NOTE: some of these are also used by catcache operations, without
 260  * any direct connection to hash indexes.  Also, the common hash_any
 261  * routine is also used by dynahash tables.
 262  */
 263 extern Datum hashchar(PG_FUNCTION_ARGS);
 264 extern Datum hashint2(PG_FUNCTION_ARGS);
 265 extern Datum hashint4(PG_FUNCTION_ARGS);
 266 extern Datum hashint8(PG_FUNCTION_ARGS);
 267 extern Datum hashoid(PG_FUNCTION_ARGS);
 268 extern Datum hashenum(PG_FUNCTION_ARGS);
 269 extern Datum hashfloat4(PG_FUNCTION_ARGS);
 270 extern Datum hashfloat8(PG_FUNCTION_ARGS);
 271 extern Datum hashoidvector(PG_FUNCTION_ARGS);
 272 extern Datum hashint2vector(PG_FUNCTION_ARGS);
 273 extern Datum hashname(PG_FUNCTION_ARGS);
 274 extern Datum hashtext(PG_FUNCTION_ARGS);
 275 extern Datum hashvarlena(PG_FUNCTION_ARGS);
 276 extern Datum hash_any(register const unsigned char *k, register int keylen);
 277 extern Datum hash_uint32(uint32 k);
 278
 279 /* private routines */
 280
 281 /* hashinsert.c */
 282 extern void _hash_doinsert(Relation rel, IndexTuple itup);
 283
 284 /* hashovfl.c */
 285 extern Buffer _hash_addovflpage(Relation rel, Buffer metabuf, Buffer buf);
 286 extern BlockNumber _hash_freeovflpage(Relation rel, Buffer ovflbuf,
 287                                    BufferAccessStrategy bstrategy);
 288 extern void _hash_initbitmap(Relation rel, HashMetaPage metap,
 289                                  BlockNumber blkno);
 290 extern void _hash_squeezebucket(Relation rel,
 291                                         Bucket bucket, BlockNumber bucket_blkno,
 292                                         BufferAccessStrategy bstrategy);
 293
 294 /* hashpage.c */
 295 extern void _hash_getlock(Relation rel, BlockNumber whichlock, int access);
 296 extern bool _hash_try_getlock(Relation rel, BlockNumber whichlock, int access);
 297 extern void _hash_droplock(Relation rel, BlockNumber whichlock, int access);
 298 extern Buffer _hash_getbuf(Relation rel, BlockNumber blkno,
 299                          int access, int flags);
 300 extern Buffer _hash_getinitbuf(Relation rel, BlockNumber blkno);
 301 extern Buffer _hash_getnewbuf(Relation rel, BlockNumber blkno);
 302 extern Buffer _hash_getbuf_with_strategy(Relation rel, BlockNumber blkno,
 303                                                    int access, int flags,
 304                                                    BufferAccessStrategy bstrategy);
 305 extern void _hash_relbuf(Relation rel, Buffer buf);
 306 extern void _hash_dropbuf(Relation rel, Buffer buf);
 307 extern void _hash_wrtbuf(Relation rel, Buffer buf);
 308 extern void _hash_chgbufaccess(Relation rel, Buffer buf, int from_access,
 309                                    int to_access);
 310 extern uint32 _hash_metapinit(Relation rel, double num_tuples);
 311 extern void _hash_pageinit(Page page, Size size);
 312 extern void _hash_expandtable(Relation rel, Buffer metabuf);
 313
 314 /* hashscan.c */
 315 extern void _hash_regscan(IndexScanDesc scan);
 316 extern void _hash_dropscan(IndexScanDesc scan);
 317 extern bool _hash_has_active_scan(Relation rel, Bucket bucket);
 318 extern void ReleaseResources_hash(void);
 319
 320 /* hashsearch.c */
 321 extern bool _hash_next(IndexScanDesc scan, ScanDirection dir);
 322 extern bool _hash_first(IndexScanDesc scan, ScanDirection dir);
 323 extern bool _hash_step(IndexScanDesc scan, Buffer *bufP, ScanDirection dir);
 324
 325 /* hashsort.c */
 326 typedef struct HSpool HSpool;   /* opaque struct in hashsort.c */
 327
 328 extern HSpool *_h_spoolinit(Relation index, uint32 num_buckets);
 329 extern void _h_spooldestroy(HSpool *hspool);
 330 extern void _h_spool(IndexTuple itup, HSpool *hspool);
 331 extern void _h_indexbuild(HSpool *hspool);
 332
 333 /* hashutil.c */
 334 extern bool _hash_checkqual(IndexScanDesc scan, IndexTuple itup);
 335 extern uint32 _hash_datum2hashkey(Relation rel, Datum key);
 336 extern uint32 _hash_datum2hashkey_type(Relation rel, Datum key, Oid keytype);
 337 extern Bucket _hash_hashkey2bucket(uint32 hashkey, uint32 maxbucket,
 338                                          uint32 highmask, uint32 lowmask);
 339 extern uint32 _hash_log2(uint32 num);
 340 extern void _hash_checkpage(Relation rel, Buffer buf, int flags);
 341 extern uint32 _hash_get_indextuple_hashkey(IndexTuple itup);
 342 extern IndexTuple _hash_form_tuple(Relation index,
 343                                  Datum *values, bool *isnull);
 344 extern OffsetNumber _hash_binsearch(Page page, uint32 hash_value);
 345 extern OffsetNumber _hash_binsearch_last(Page page, uint32 hash_value);
 346
 347 /* hash.c */
 348 extern void hash_redo(XLogRecPtr lsn, XLogRecord *record);
 349 extern void hash_desc(StringInfo buf, uint8 xl_info, char *rec);
 350
 351 #endif   /* HASH_H */