dynahash.c

来自「postgresql8.3.4源码,开源数据库」· C语言 代码 · 共 1,529 行 · 第 1/3 页

/*-------------------------------------------------------------------------
 *
 * dynahash.c
 *	  dynamic hash tables
 *
 * dynahash.c supports both local-to-a-backend hash tables and hash tables in
 * shared memory.  For shared hash tables, it is the caller's responsibility
 * to provide appropriate access interlocking.	The simplest convention is
 * that a single LWLock protects the whole hash table.	Searches (HASH_FIND or
 * hash_seq_search) need only shared lock, but any update requires exclusive
 * lock.  For heavily-used shared tables, the single-lock approach creates a
 * concurrency bottleneck, so we also support "partitioned" locking wherein
 * there are multiple LWLocks guarding distinct subsets of the table.  To use
 * a hash table in partitioned mode, the HASH_PARTITION flag must be given
 * to hash_create.	This prevents any attempt to split buckets on-the-fly.
 * Therefore, each hash bucket chain operates independently, and no fields
 * of the hash header change after init except nentries and freeList.
 * A partitioned table uses a spinlock to guard changes of those two fields.
 * This lets any subset of the hash buckets be treated as a separately
 * lockable partition.	We expect callers to use the low-order bits of a
 * lookup key's hash value as a partition number --- this will work because
 * of the way calc_bucket() maps hash values to bucket numbers.
 *
 * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
 *	  $PostgreSQL: pgsql/src/backend/utils/hash/dynahash.c,v 1.78 2008/01/01 19:45:53 momjian Exp $
 *
 *-------------------------------------------------------------------------
 */

/*
 * Original comments:
 *
 * Dynamic hashing, after CACM April 1988 pp 446-457, by Per-Ake Larson.
 * Coded into C, with minor code improvements, and with hsearch(3) interface,
 * by [e-mail], Jul 26, 1988: 13:16;
 * also, hcreate/hdestroy routines added to simulate hsearch(3).
 *
 * These routines simulate hsearch(3) and family, with the important
 * difference that the hash table is dynamic - can grow indefinitely
 * beyond its original size (as supplied to hcreate()).
 *
 * Performance appears to be comparable to that of hsearch(3).
 * The 'source-code' options referred to in hsearch(3)'s 'man' page
 * are not implemented; otherwise functionality is identical.
 *
 * Compilation controls:
 * DEBUG controls some informative traces, mainly for debugging.
 * HASH_STATISTICS causes HashAccesses and HashCollisions to be maintained;
 * when combined with HASH_DEBUG, these are displayed by hdestroy().
 *
 * Problems & fixes to [e-mail]. WARNING: relies on pre-processor
 * concatenation property, in probably unnecessary code 'optimisation'.
 *
 * Modified [e-mail] February 1990
 *		added multiple table interface
 * Modified by [e-mail] April 1990
 *		changed ctl structure for shared memory
 */

#include "postgres.h"

#include "access/xact.h"
#include "storage/shmem.h"
#include "storage/spin.h"
#include "utils/dynahash.h"
#include "utils/memutils.h"


/*
 * Constants
 *
 * A hash table has a top-level "directory", each of whose entries points
 * to a "segment" of ssize bucket headers.	The maximum number of hash
 * buckets is thus dsize * ssize (but dsize may be expansible).  Of course,
 * the number of records in the table can be larger, but we don't want a
 * whole lot of records per bucket or performance goes down.
 *
 * In a hash table allocated in shared memory, the directory cannot be
 * expanded because it must stay at a fixed address.  The directory size
 * should be selected using hash_select_dirsize (and you'd better have
 * a good idea of the maximum number of entries!).	For non-shared hash
 * tables, the initial directory size can be left at the default.
 */
#define DEF_SEGSIZE			   256
#define DEF_SEGSIZE_SHIFT	   8	/* must be log2(DEF_SEGSIZE) */
#define DEF_DIRSIZE			   256
#define DEF_FFACTOR			   1	/* default fill factor */


/* A hash bucket is a linked list of HASHELEMENTs */
typedef HASHELEMENT *HASHBUCKET;

/* A hash segment is an array of bucket headers */
typedef HASHBUCKET *HASHSEGMENT;

/*
 * Header structure for a hash table --- contains all changeable info
 *
 * In a shared-memory hash table, the HASHHDR is in shared memory, while
 * each backend has a local HTAB struct.  For a non-shared table, there isn't
 * any functional difference between HASHHDR and HTAB, but we separate them
 * anyway to share code between shared and non-shared tables.
 */
struct HASHHDR
{
	/* In a partitioned table, take this lock to touch nentries or freeList */
	slock_t		mutex;			/* unused if not partitioned table */

	/* These fields change during entry addition/deletion */
	long		nentries;		/* number of entries in hash table */
	HASHELEMENT *freeList;		/* linked list of free elements */

	/* These fields can change, but not in a partitioned table */
	/* Also, dsize can't change in a shared table, even if unpartitioned */
	long		dsize;			/* directory size */
	long		nsegs;			/* number of allocated segments (<= dsize) */
	uint32		max_bucket;		/* ID of maximum bucket in use */
	uint32		high_mask;		/* mask to modulo into entire table */
	uint32		low_mask;		/* mask to modulo into lower half of table */

	/* These fields are fixed at hashtable creation */
	Size		keysize;		/* hash key length in bytes */
	Size		entrysize;		/* total user element size in bytes */
	long		num_partitions; /* # partitions (must be power of 2), or 0 */
	long		ffactor;		/* target fill factor */
	long		max_dsize;		/* 'dsize' limit if directory is fixed size */
	long		ssize;			/* segment size --- must be power of 2 */
	int			sshift;			/* segment shift = log2(ssize) */
	int			nelem_alloc;	/* number of entries to allocate at once */

#ifdef HASH_STATISTICS

	/*
	 * Count statistics here.  NB: stats code doesn't bother with mutex, so
	 * counts could be corrupted a bit in a partitioned table.
	 */
	long		accesses;
	long		collisions;
#endif
};

#define IS_PARTITIONED(hctl)  ((hctl)->num_partitions != 0)

/*
 * Top control structure for a hashtable --- in a shared table, each backend
 * has its own copy (OK since no fields change at runtime)
 */
struct HTAB
{
	HASHHDR    *hctl;			/* => shared control information */
	HASHSEGMENT *dir;			/* directory of segment starts */
	HashValueFunc hash;			/* hash function */
	HashCompareFunc match;		/* key comparison function */
	HashCopyFunc keycopy;		/* key copying function */
	HashAllocFunc alloc;		/* memory allocator */
	MemoryContext hcxt;			/* memory context if default allocator used */
	char	   *tabname;		/* table name (for error messages) */
	bool		isshared;		/* true if table is in shared memory */

	/* freezing a shared table isn't allowed, so we can keep state here */
	bool		frozen;			/* true = no more inserts allowed */

	/* We keep local copies of these fixed values to reduce contention */
	Size		keysize;		/* hash key length in bytes */
	long		ssize;			/* segment size --- must be power of 2 */
	int			sshift;			/* segment shift = log2(ssize) */
};

/*
 * Key (also entry) part of a HASHELEMENT
 */
#define ELEMENTKEY(helem)  (((char *)(helem)) + MAXALIGN(sizeof(HASHELEMENT)))

/*
 * Fast MOD arithmetic, assuming that y is a power of 2 !
 */
#define MOD(x,y)			   ((x) & ((y)-1))

#if HASH_STATISTICS
static long hash_accesses,
			hash_collisions,
			hash_expansions;
#endif

/*
 * Private function prototypes
 */
static void *DynaHashAlloc(Size size);
static HASHSEGMENT seg_alloc(HTAB *hashp);
static bool element_alloc(HTAB *hashp, int nelem);
static bool dir_realloc(HTAB *hashp);
static bool expand_table(HTAB *hashp);
static HASHBUCKET get_hash_entry(HTAB *hashp);
static void hdefault(HTAB *hashp);
static int	choose_nelem_alloc(Size entrysize);
static bool init_htab(HTAB *hashp, long nelem);
static void hash_corrupted(HTAB *hashp);
static void register_seq_scan(HTAB *hashp);
static void deregister_seq_scan(HTAB *hashp);
static bool has_seq_scans(HTAB *hashp);


/*
 * memory allocation support
 */
static MemoryContext CurrentDynaHashCxt = NULL;

static void *
DynaHashAlloc(Size size)
{
	Assert(MemoryContextIsValid(CurrentDynaHashCxt));
	return MemoryContextAlloc(CurrentDynaHashCxt, size);
}


/*
 * HashCompareFunc for string keys
 *
 * Because we copy keys with strlcpy(), they will be truncated at keysize-1
 * bytes, so we can only compare that many ... hence strncmp is almost but
 * not quite the right thing.
 */
static int
string_compare(const char *key1, const char *key2, Size keysize)
{
	return strncmp(key1, key2, keysize - 1);
}


/************************** CREATE ROUTINES **********************/

/*
 * hash_create -- create a new dynamic hash table
 *
 *	tabname: a name for the table (for debugging purposes)
 *	nelem: maximum number of elements expected
 *	*info: additional table parameters, as indicated by flags
 *	flags: bitmask indicating which parameters to take from *info
 *
 * Note: for a shared-memory hashtable, nelem needs to be a pretty good
 * estimate, since we can't expand the table on the fly.  But an unshared
 * hashtable can be expanded on-the-fly, so it's better for nelem to be
 * on the small side and let the table grow if it's exceeded.  An overly
 * large nelem will penalize hash_seq_search speed without buying much.
 */
HTAB *
hash_create(const char *tabname, long nelem, HASHCTL *info, int flags)
{
	HTAB	   *hashp;
	HASHHDR    *hctl;

	/*
	 * For shared hash tables, we have a local hash header (HTAB struct) that
	 * we allocate in TopMemoryContext; all else is in shared memory.
	 *
	 * For non-shared hash tables, everything including the hash header is in
	 * a memory context created specially for the hash table --- this makes
	 * hash_destroy very simple.  The memory context is made a child of either
	 * a context specified by the caller, or TopMemoryContext if nothing is
	 * specified.
	 */
	if (flags & HASH_SHARED_MEM)
	{
		/* Set up to allocate the hash header */
		CurrentDynaHashCxt = TopMemoryContext;
	}
	else
	{
		/* Create the hash table's private memory context */
		if (flags & HASH_CONTEXT)
			CurrentDynaHashCxt = info->hcxt;
		else
			CurrentDynaHashCxt = TopMemoryContext;
		CurrentDynaHashCxt = AllocSetContextCreate(CurrentDynaHashCxt,
												   tabname,
												   ALLOCSET_DEFAULT_MINSIZE,
												   ALLOCSET_DEFAULT_INITSIZE,
												   ALLOCSET_DEFAULT_MAXSIZE);
	}

	/* Initialize the hash header, plus a copy of the table name */
	hashp = (HTAB *) DynaHashAlloc(sizeof(HTAB) + strlen(tabname) +1);
	MemSet(hashp, 0, sizeof(HTAB));

	hashp->tabname = (char *) (hashp + 1);
	strcpy(hashp->tabname, tabname);

	if (flags & HASH_FUNCTION)
		hashp->hash = info->hash;
	else
		hashp->hash = string_hash;		/* default hash function */

	/*
	 * If you don't specify a match function, it defaults to string_compare if
	 * you used string_hash (either explicitly or by default) and to memcmp
	 * otherwise.  (Prior to PostgreSQL 7.4, memcmp was always used.)
	 */
	if (flags & HASH_COMPARE)
		hashp->match = info->match;
	else if (hashp->hash == string_hash)
		hashp->match = (HashCompareFunc) string_compare;
	else
		hashp->match = memcmp;

	/*
	 * Similarly, the key-copying function defaults to strlcpy or memcpy.
	 */
	if (flags & HASH_KEYCOPY)
		hashp->keycopy = info->keycopy;
	else if (hashp->hash == string_hash)
		hashp->keycopy = (HashCopyFunc) strlcpy;
	else
		hashp->keycopy = memcpy;

	if (flags & HASH_ALLOC)
		hashp->alloc = info->alloc;
	else
		hashp->alloc = DynaHashAlloc;

	if (flags & HASH_SHARED_MEM)
	{
		/*
		 * ctl structure and directory are preallocated for shared memory
		 * tables.	Note that HASH_DIRSIZE and HASH_ALLOC had better be set as
		 * well.
		 */
		hashp->hctl = info->hctl;
		hashp->dir = (HASHSEGMENT *) (((char *) info->hctl) + sizeof(HASHHDR));
		hashp->hcxt = NULL;
		hashp->isshared = true;

		/* hash table already exists, we're just attaching to it */
		if (flags & HASH_ATTACH)
		{
			/* make local copies of some heavily-used values */
			hctl = hashp->hctl;
			hashp->keysize = hctl->keysize;
			hashp->ssize = hctl->ssize;
			hashp->sshift = hctl->sshift;

			return hashp;
		}
	}
	else
	{
		/* setup hash table defaults */
		hashp->hctl = NULL;
		hashp->dir = NULL;
		hashp->hcxt = CurrentDynaHashCxt;
		hashp->isshared = false;
	}

	if (!hashp->hctl)
	{
		hashp->hctl = (HASHHDR *) hashp->alloc(sizeof(HASHHDR));
		if (!hashp->hctl)
			ereport(ERROR,
					(errcode(ERRCODE_OUT_OF_MEMORY),
					 errmsg("out of memory")));
	}

	hashp->frozen = false;

	hdefault(hashp);

	hctl = hashp->hctl;

	if (flags & HASH_PARTITION)
	{
		/* Doesn't make sense to partition a local hash table */
		Assert(flags & HASH_SHARED_MEM);
		/* # of partitions had better be a power of 2 */
		Assert(info->num_partitions == (1L << my_log2(info->num_partitions)));

		hctl->num_partitions = info->num_partitions;
	}

	if (flags & HASH_SEGMENT)
	{
		hctl->ssize = info->ssize;
		hctl->sshift = my_log2(info->ssize);
		/* ssize had better be a power of 2 */
		Assert(hctl->ssize == (1L << hctl->sshift));
	}
	if (flags & HASH_FFACTOR)
		hctl->ffactor = info->ffactor;

	/*
	 * SHM hash tables have fixed directory size passed by the caller.
	 */
	if (flags & HASH_DIRSIZE)
	{
		hctl->max_dsize = info->max_dsize;
		hctl->dsize = info->dsize;
	}

	/*
	 * hash table now allocates space for key and data but you have to say how
	 * much space to allocate
	 */
	if (flags & HASH_ELEM)
	{
		Assert(info->entrysize >= info->keysize);
		hctl->keysize = info->keysize;
		hctl->entrysize = info->entrysize;
	}

	/* make local copies of heavily-used constant fields */
	hashp->keysize = hctl->keysize;
	hashp->ssize = hctl->ssize;
	hashp->sshift = hctl->sshift;

	/* Build the hash directory structure */
	if (!init_htab(hashp, nelem))
		elog(ERROR, "failed to initialize hash table \"%s\"", hashp->tabname);

	/*
	 * For a shared hash table, preallocate the requested number of elements.
	 * This reduces problems with run-time out-of-shared-memory conditions.
	 *
	 * For a non-shared hash table, preallocate the requested number of
	 * elements if it's less than our chosen nelem_alloc.  This avoids wasting
	 * space if the caller correctly estimates a small table size.
	 */
	if ((flags & HASH_SHARED_MEM) ||
		nelem < hctl->nelem_alloc)
	{
		if (!element_alloc(hashp, (int) nelem))
			ereport(ERROR,
					(errcode(ERRCODE_OUT_OF_MEMORY),
					 errmsg("out of memory")));
	}

	return hashp;
}

/*
 * Set default HASHHDR parameters.
 */
static void
hdefault(HTAB *hashp)
{
	HASHHDR    *hctl = hashp->hctl;

	MemSet(hctl, 0, sizeof(HASHHDR));

	hctl->nentries = 0;
	hctl->freeList = NULL;

	hctl->dsize = DEF_DIRSIZE;
	hctl->nsegs = 0;

	/* rather pointless defaults for key & entry size */
	hctl->keysize = sizeof(char *);
	hctl->entrysize = 2 * sizeof(char *);

	hctl->num_partitions = 0;	/* not partitioned */

	hctl->ffactor = DEF_FFACTOR;

	/* table has no fixed maximum size */
	hctl->max_dsize = NO_MAX_DSIZE;

	hctl->ssize = DEF_SEGSIZE;
	hctl->sshift = DEF_SEGSIZE_SHIFT;

#ifdef HASH_STATISTICS
	hctl->accesses = hctl->collisions = 0;
#endif
}

/*
 * Given the user-specified entry size, choose nelem_alloc, ie, how many
 * elements to add to the hash table when we need more.
 */
static int
choose_nelem_alloc(Size entrysize)
{
	int			nelem_alloc;
	Size		elementSize;
	Size		allocSize;

	/* Each element has a HASHELEMENT header plus user data. */
	/* NB: this had better match element_alloc() */
	elementSize = MAXALIGN(sizeof(HASHELEMENT)) + MAXALIGN(entrysize);

	/*
	 * The idea here is to choose nelem_alloc at least 32, but round up so
	 * that the allocation request will be a power of 2 or just less. This
	 * makes little difference for hash tables in shared memory, but for hash
	 * tables managed by palloc, the allocation request will be rounded up to
	 * a power of 2 anyway.  If we fail to take this into account, we'll waste
	 * as much as half the allocated space.
	 */
	allocSize = 32 * 4;			/* assume elementSize at least 8 */
	do
	{
		allocSize <<= 1;
		nelem_alloc = allocSize / elementSize;
	} while (nelem_alloc < 32);

	return nelem_alloc;
}

/*
 * Compute derived fields of hctl and build the initial directory/segment