freebsd-nq/lib/libc/db/hash/hash.h
Andriy Gapon f3c733e82b libc/db/hash: cap auto-tuned block size with a value that actually works
This fix mostly matters after r206129 that made it possible for
st_blksize to be greater than 4K.  For this reason, this change should
be MFC-ed before r206129.
Also, it seems that all FreeBSD uitlities that use db(3) hash databases
and create new databases in files, specify their own block size value
and thus do not depend on block size autotuning.

PR:		bin/144446
Submitted by:	Peter Jeremy <peterjeremy@acm.org>
MFC after:	5 days
2010-04-05 10:12:21 +00:00

291 lines
9.8 KiB
C

/*-
* Copyright (c) 1990, 1993, 1994
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Margo Seltzer.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)hash.h 8.3 (Berkeley) 5/31/94
* $FreeBSD$
*/
/* Operations */
typedef enum {
HASH_GET, HASH_PUT, HASH_PUTNEW, HASH_DELETE, HASH_FIRST, HASH_NEXT
} ACTION;
/* Buffer Management structures */
typedef struct _bufhead BUFHEAD;
struct _bufhead {
BUFHEAD *prev; /* LRU links */
BUFHEAD *next; /* LRU links */
BUFHEAD *ovfl; /* Overflow page buffer header */
u_int32_t addr; /* Address of this page */
char *page; /* Actual page data */
char flags;
#define BUF_MOD 0x0001
#define BUF_DISK 0x0002
#define BUF_BUCKET 0x0004
#define BUF_PIN 0x0008
};
#define IS_BUCKET(X) ((X) & BUF_BUCKET)
typedef BUFHEAD **SEGMENT;
/* Hash Table Information */
typedef struct hashhdr { /* Disk resident portion */
int32_t magic; /* Magic NO for hash tables */
int32_t version; /* Version ID */
u_int32_t lorder; /* Byte Order */
int32_t bsize; /* Bucket/Page Size */
int32_t bshift; /* Bucket shift */
int32_t dsize; /* Directory Size */
int32_t ssize; /* Segment Size */
int32_t sshift; /* Segment shift */
int32_t ovfl_point; /* Where overflow pages are being
* allocated */
int32_t last_freed; /* Last overflow page freed */
u_int32_t max_bucket; /* ID of Maximum bucket in use */
u_int32_t high_mask; /* Mask to modulo into entire table */
u_int32_t low_mask; /* Mask to modulo into lower half of
* table */
u_int32_t ffactor; /* Fill factor */
int32_t nkeys; /* Number of keys in hash table */
int32_t hdrpages; /* Size of table header */
int32_t h_charkey; /* value of hash(CHARKEY) */
#define NCACHED 32 /* number of bit maps and spare
* points */
int32_t spares[NCACHED];/* spare pages for overflow */
u_int16_t bitmaps[NCACHED]; /* address of overflow page
* bitmaps */
} HASHHDR;
typedef struct htab { /* Memory resident data structure */
HASHHDR hdr; /* Header */
int nsegs; /* Number of allocated segments */
int exsegs; /* Number of extra allocated
* segments */
u_int32_t /* Hash function */
(*hash)(const void *, size_t);
int flags; /* Flag values */
int fp; /* File pointer */
char *tmp_buf; /* Temporary Buffer for BIG data */
char *tmp_key; /* Temporary Buffer for BIG keys */
BUFHEAD *cpage; /* Current page */
int cbucket; /* Current bucket */
int cndx; /* Index of next item on cpage */
int error; /* Error Number -- for DBM
* compatibility */
int new_file; /* Indicates if fd is backing store
* or no */
int save_file; /* Indicates whether we need to flush
* file at
* exit */
u_int32_t *mapp[NCACHED]; /* Pointers to page maps */
int nmaps; /* Initial number of bitmaps */
int nbufs; /* Number of buffers left to
* allocate */
BUFHEAD bufhead; /* Header of buffer lru list */
SEGMENT *dir; /* Hash Bucket directory */
} HTAB;
/*
* Constants
*/
#define MAX_BSIZE 32768 /* 2^15 but should be 65536 */
#define MIN_BUFFERS 6
#define MINHDRSIZE 512
#define DEF_BUFSIZE 65536 /* 64 K */
#define DEF_BUCKET_SIZE 4096
#define DEF_BUCKET_SHIFT 12 /* log2(BUCKET) */
#define DEF_SEGSIZE 256
#define DEF_SEGSIZE_SHIFT 8 /* log2(SEGSIZE) */
#define DEF_DIRSIZE 256
#define DEF_FFACTOR 65536
#define MIN_FFACTOR 4
#define SPLTMAX 8
#define CHARKEY "%$sniglet^&"
#define NUMKEY 1038583
#define BYTE_SHIFT 3
#define INT_TO_BYTE 2
#define INT_BYTE_SHIFT 5
#define ALL_SET ((u_int32_t)0xFFFFFFFF)
#define ALL_CLEAR 0
#define PTROF(X) ((BUFHEAD *)((ptrdiff_t)(X)&~0x3))
#define ISMOD(X) ((u_int32_t)(ptrdiff_t)(X)&0x1)
#define DOMOD(X) ((X) = (char *)((ptrdiff_t)(X)|0x1))
#define ISDISK(X) ((u_int32_t)(ptrdiff_t)(X)&0x2)
#define DODISK(X) ((X) = (char *)((ptrdiff_t)(X)|0x2))
#define BITS_PER_MAP 32
/* Given the address of the beginning of a big map, clear/set the nth bit */
#define CLRBIT(A, N) ((A)[(N)/BITS_PER_MAP] &= ~(1<<((N)%BITS_PER_MAP)))
#define SETBIT(A, N) ((A)[(N)/BITS_PER_MAP] |= (1<<((N)%BITS_PER_MAP)))
#define ISSET(A, N) ((A)[(N)/BITS_PER_MAP] & (1<<((N)%BITS_PER_MAP)))
/* Overflow management */
/*
* Overflow page numbers are allocated per split point. At each doubling of
* the table, we can allocate extra pages. So, an overflow page number has
* the top 5 bits indicate which split point and the lower 11 bits indicate
* which page at that split point is indicated (pages within split points are
* numberered starting with 1).
*/
#define SPLITSHIFT 11
#define SPLITMASK 0x7FF
#define SPLITNUM(N) (((u_int32_t)(N)) >> SPLITSHIFT)
#define OPAGENUM(N) ((N) & SPLITMASK)
#define OADDR_OF(S,O) ((u_int32_t)((u_int32_t)(S) << SPLITSHIFT) + (O))
#define BUCKET_TO_PAGE(B) \
(B) + hashp->HDRPAGES + ((B) ? hashp->SPARES[__log2((B)+1)-1] : 0)
#define OADDR_TO_PAGE(B) \
BUCKET_TO_PAGE ( (1 << SPLITNUM((B))) -1 ) + OPAGENUM((B));
/*
* page.h contains a detailed description of the page format.
*
* Normally, keys and data are accessed from offset tables in the top of
* each page which point to the beginning of the key and data. There are
* four flag values which may be stored in these offset tables which indicate
* the following:
*
*
* OVFLPAGE Rather than a key data pair, this pair contains
* the address of an overflow page. The format of
* the pair is:
* OVERFLOW_PAGE_NUMBER OVFLPAGE
*
* PARTIAL_KEY This must be the first key/data pair on a page
* and implies that page contains only a partial key.
* That is, the key is too big to fit on a single page
* so it starts on this page and continues on the next.
* The format of the page is:
* KEY_OFF PARTIAL_KEY OVFL_PAGENO OVFLPAGE
*
* KEY_OFF -- offset of the beginning of the key
* PARTIAL_KEY -- 1
* OVFL_PAGENO - page number of the next overflow page
* OVFLPAGE -- 0
*
* FULL_KEY This must be the first key/data pair on the page. It
* is used in two cases.
*
* Case 1:
* There is a complete key on the page but no data
* (because it wouldn't fit). The next page contains
* the data.
*
* Page format it:
* KEY_OFF FULL_KEY OVFL_PAGENO OVFL_PAGE
*
* KEY_OFF -- offset of the beginning of the key
* FULL_KEY -- 2
* OVFL_PAGENO - page number of the next overflow page
* OVFLPAGE -- 0
*
* Case 2:
* This page contains no key, but part of a large
* data field, which is continued on the next page.
*
* Page format it:
* DATA_OFF FULL_KEY OVFL_PAGENO OVFL_PAGE
*
* KEY_OFF -- offset of the beginning of the data on
* this page
* FULL_KEY -- 2
* OVFL_PAGENO - page number of the next overflow page
* OVFLPAGE -- 0
*
* FULL_KEY_DATA
* This must be the first key/data pair on the page.
* There are two cases:
*
* Case 1:
* This page contains a key and the beginning of the
* data field, but the data field is continued on the
* next page.
*
* Page format is:
* KEY_OFF FULL_KEY_DATA OVFL_PAGENO DATA_OFF
*
* KEY_OFF -- offset of the beginning of the key
* FULL_KEY_DATA -- 3
* OVFL_PAGENO - page number of the next overflow page
* DATA_OFF -- offset of the beginning of the data
*
* Case 2:
* This page contains the last page of a big data pair.
* There is no key, only the tail end of the data
* on this page.
*
* Page format is:
* DATA_OFF FULL_KEY_DATA <OVFL_PAGENO> <OVFLPAGE>
*
* DATA_OFF -- offset of the beginning of the data on
* this page
* FULL_KEY_DATA -- 3
* OVFL_PAGENO - page number of the next overflow page
* OVFLPAGE -- 0
*
* OVFL_PAGENO and OVFLPAGE are optional (they are
* not present if there is no next page).
*/
#define OVFLPAGE 0
#define PARTIAL_KEY 1
#define FULL_KEY 2
#define FULL_KEY_DATA 3
#define REAL_KEY 4
/* Short hands for accessing structure */
#define BSIZE hdr.bsize
#define BSHIFT hdr.bshift
#define DSIZE hdr.dsize
#define SGSIZE hdr.ssize
#define SSHIFT hdr.sshift
#define LORDER hdr.lorder
#define OVFL_POINT hdr.ovfl_point
#define LAST_FREED hdr.last_freed
#define MAX_BUCKET hdr.max_bucket
#define FFACTOR hdr.ffactor
#define HIGH_MASK hdr.high_mask
#define LOW_MASK hdr.low_mask
#define NKEYS hdr.nkeys
#define HDRPAGES hdr.hdrpages
#define SPARES hdr.spares
#define BITMAPS hdr.bitmaps
#define VERSION hdr.version
#define MAGIC hdr.magic
#define NEXT_FREE hdr.next_free
#define H_CHARKEY hdr.h_charkey