/*
 *	_hash_getovflpage()
 *
 *	Find an available overflow page and return it.	The returned buffer
 *	is pinned and write-locked, and has had _hash_pageinit() applied,
 *	but it is caller's responsibility to fill the special space.
 *
 * The caller must hold a pin, but no lock, on the metapage buffer.
 * That buffer is left in the same state at exit.
 */
static Buffer
_hash_getovflpage(Relation rel, Buffer metabuf)
{
    HashMetaPage metap;
    Buffer		mapbuf = 0;
    Buffer		newbuf;
    BlockNumber blkno;
    uint32		orig_firstfree;
    uint32		splitnum;
    uint32	   *freep = NULL;
    uint32		max_ovflpg;
    uint32		bit;
    uint32		first_page;
    uint32		last_bit;
    uint32		last_page;
    uint32		i,
                j;

    /* Get exclusive lock on the meta page */
    _hash_chgbufaccess(rel, metabuf, HASH_NOLOCK, HASH_WRITE);

    _hash_checkpage(rel, metabuf, LH_META_PAGE);
    metap = HashPageGetMeta(BufferGetPage(metabuf));

    /* start search at hashm_firstfree */
    orig_firstfree = metap->hashm_firstfree;
    first_page = orig_firstfree >> BMPG_SHIFT(metap);
    bit = orig_firstfree & BMPG_MASK(metap);
    i = first_page;
    j = bit / BITS_PER_MAP;
    bit &= ~(BITS_PER_MAP - 1);

    /* outer loop iterates once per bitmap page */
    for (;;)
    {
        BlockNumber mapblkno;
        Page		mappage;
        uint32		last_inpage;

        /* want to end search with the last existing overflow page */
        splitnum = metap->hashm_ovflpoint;
        max_ovflpg = metap->hashm_spares[splitnum] - 1;
        last_page = max_ovflpg >> BMPG_SHIFT(metap);
        last_bit = max_ovflpg & BMPG_MASK(metap);

        if (i > last_page)
            break;

        Assert(i < metap->hashm_nmaps);
        mapblkno = metap->hashm_mapp[i];

        if (i == last_page)
            last_inpage = last_bit;
        else
            last_inpage = BMPGSZ_BIT(metap) - 1;

        /* Release exclusive lock on metapage while reading bitmap page */
        _hash_chgbufaccess(rel, metabuf, HASH_READ, HASH_NOLOCK);

        mapbuf = _hash_getbuf(rel, mapblkno, HASH_WRITE, LH_BITMAP_PAGE);
        mappage = BufferGetPage(mapbuf);
        freep = HashPageGetBitmap(mappage);

        for (; bit <= last_inpage; j++, bit += BITS_PER_MAP)
        {
            if (freep[j] != ALL_SET)
                goto found;
        }

        /* No free space here, try to advance to next map page */
        _hash_relbuf(rel, mapbuf);
        i++;
        j = 0;					/* scan from start of next map page */
        bit = 0;

        /* Reacquire exclusive lock on the meta page */
        _hash_chgbufaccess(rel, metabuf, HASH_NOLOCK, HASH_WRITE);
    }

    /*
     * No free pages --- have to extend the relation to add an overflow page.
     * First, check to see if we have to add a new bitmap page too.
     */
    if (last_bit == (uint32) (BMPGSZ_BIT(metap) - 1))
    {
        /*
         * We create the new bitmap page with all pages marked "in use".
         * Actually two pages in the new bitmap's range will exist
         * immediately: the bitmap page itself, and the following page which
         * is the one we return to the caller.	Both of these are correctly
         * marked "in use".  Subsequent pages do not exist yet, but it is
         * convenient to pre-mark them as "in use" too.
         */
        bit = metap->hashm_spares[splitnum];
        _hash_initbitmap(rel, metap, bitno_to_blkno(metap, bit), MAIN_FORKNUM);
        metap->hashm_spares[splitnum]++;
    }
    else
    {
        /*
         * Nothing to do here; since the page will be past the last used page,
         * we know its bitmap bit was preinitialized to "in use".
         */
    }

    /* Calculate address of the new overflow page */
    bit = metap->hashm_spares[splitnum];
    blkno = bitno_to_blkno(metap, bit);

    /*
     * Fetch the page with _hash_getnewbuf to ensure smgr's idea of the
     * relation length stays in sync with ours.  XXX It's annoying to do this
     * with metapage write lock held; would be better to use a lock that
     * doesn't block incoming searches.
     */
    newbuf = _hash_getnewbuf(rel, blkno, MAIN_FORKNUM);

    metap->hashm_spares[splitnum]++;

    /*
     * Adjust hashm_firstfree to avoid redundant searches.	But don't risk
     * changing it if someone moved it while we were searching bitmap pages.
     */
    if (metap->hashm_firstfree == orig_firstfree)
        metap->hashm_firstfree = bit + 1;

    /* Write updated metapage and release lock, but not pin */
    _hash_chgbufaccess(rel, metabuf, HASH_WRITE, HASH_NOLOCK);

    return newbuf;

found:
    /* convert bit to bit number within page */
    bit += _hash_firstfreebit(freep[j]);

    /* mark page "in use" in the bitmap */
    SETBIT(freep, bit);
    _hash_wrtbuf(rel, mapbuf);

    /* Reacquire exclusive lock on the meta page */
    _hash_chgbufaccess(rel, metabuf, HASH_NOLOCK, HASH_WRITE);

    /* convert bit to absolute bit number */
    bit += (i << BMPG_SHIFT(metap));

    /* Calculate address of the recycled overflow page */
    blkno = bitno_to_blkno(metap, bit);

    /*
     * Adjust hashm_firstfree to avoid redundant searches.	But don't risk
     * changing it if someone moved it while we were searching bitmap pages.
     */
    if (metap->hashm_firstfree == orig_firstfree)
    {
        metap->hashm_firstfree = bit + 1;

        /* Write updated metapage and release lock, but not pin */
        _hash_chgbufaccess(rel, metabuf, HASH_WRITE, HASH_NOLOCK);
    }
    else
    {
        /* We didn't change the metapage, so no need to write */
        _hash_chgbufaccess(rel, metabuf, HASH_READ, HASH_NOLOCK);
    }

    /* Fetch, init, and return the recycled page */
    return _hash_getinitbuf(rel, blkno);
}
/*
 *	_hash_freeovflpage() -
 *
 *	Remove this overflow page from its bucket's chain, and mark the page as
 *	free.  On entry, ovflbuf is write-locked; it is released before exiting.
 *
 *	Since this function is invoked in VACUUM, we provide an access strategy
 *	parameter that controls fetches of the bucket pages.
 *
 *	Returns the block number of the page that followed the given page
 *	in the bucket, or InvalidBlockNumber if no following page.
 *
 *	NB: caller must not hold lock on metapage, nor on either page that's
 *	adjacent in the bucket chain.  The caller had better hold exclusive lock
 *	on the bucket, too.
 */
BlockNumber
_hash_freeovflpage(Relation rel, Buffer ovflbuf,
                   BufferAccessStrategy bstrategy)
{
    HashMetaPage metap;
    Buffer		metabuf;
    Buffer		mapbuf;
    BlockNumber ovflblkno;
    BlockNumber prevblkno;
    BlockNumber blkno;
    BlockNumber nextblkno;
    HashPageOpaque ovflopaque;
    Page		ovflpage;
    Page		mappage;
    uint32	   *freep;
    uint32		ovflbitno;
    int32		bitmappage,
                bitmapbit;
    Bucket bucket PG_USED_FOR_ASSERTS_ONLY;

    /* Get information from the doomed page */
    _hash_checkpage(rel, ovflbuf, LH_OVERFLOW_PAGE);
    ovflblkno = BufferGetBlockNumber(ovflbuf);
    ovflpage = BufferGetPage(ovflbuf);
    ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
    nextblkno = ovflopaque->hasho_nextblkno;
    prevblkno = ovflopaque->hasho_prevblkno;
    bucket = ovflopaque->hasho_bucket;

    /*
     * Zero the page for debugging's sake; then write and release it. (Note:
     * if we failed to zero the page here, we'd have problems with the Assert
     * in _hash_pageinit() when the page is reused.)
     */
    MemSet(ovflpage, 0, BufferGetPageSize(ovflbuf));
    _hash_wrtbuf(rel, ovflbuf);

    /*
     * Fix up the bucket chain.  this is a doubly-linked list, so we must fix
     * up the bucket chain members behind and ahead of the overflow page being
     * deleted.  No concurrency issues since we hold exclusive lock on the
     * entire bucket.
     */
    if (BlockNumberIsValid(prevblkno))
    {
        Buffer		prevbuf = _hash_getbuf_with_strategy(rel,
                              prevblkno,
                              HASH_WRITE,
                              LH_BUCKET_PAGE | LH_OVERFLOW_PAGE,
                              bstrategy);
        Page		prevpage = BufferGetPage(prevbuf);
        HashPageOpaque prevopaque = (HashPageOpaque) PageGetSpecialPointer(prevpage);

        Assert(prevopaque->hasho_bucket == bucket);
        prevopaque->hasho_nextblkno = nextblkno;
        _hash_wrtbuf(rel, prevbuf);
    }
    if (BlockNumberIsValid(nextblkno))
    {
        Buffer		nextbuf = _hash_getbuf_with_strategy(rel,
                              nextblkno,
                              HASH_WRITE,
                              LH_OVERFLOW_PAGE,
                              bstrategy);
        Page		nextpage = BufferGetPage(nextbuf);
        HashPageOpaque nextopaque = (HashPageOpaque) PageGetSpecialPointer(nextpage);

        Assert(nextopaque->hasho_bucket == bucket);
        nextopaque->hasho_prevblkno = prevblkno;
        _hash_wrtbuf(rel, nextbuf);
    }

    /* Note: bstrategy is intentionally not used for metapage and bitmap */

    /* Read the metapage so we can determine which bitmap page to use */
    metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
    metap = HashPageGetMeta(BufferGetPage(metabuf));

    /* Identify which bit to set */
    ovflbitno = blkno_to_bitno(metap, ovflblkno);

    bitmappage = ovflbitno >> BMPG_SHIFT(metap);
    bitmapbit = ovflbitno & BMPG_MASK(metap);

    if (bitmappage >= metap->hashm_nmaps)
        elog(ERROR, "invalid overflow bit number %u", ovflbitno);
    blkno = metap->hashm_mapp[bitmappage];

    /* Release metapage lock while we access the bitmap page */
    _hash_chgbufaccess(rel, metabuf, HASH_READ, HASH_NOLOCK);

    /* Clear the bitmap bit to indicate that this overflow page is free */
    mapbuf = _hash_getbuf(rel, blkno, HASH_WRITE, LH_BITMAP_PAGE);
    mappage = BufferGetPage(mapbuf);
    freep = HashPageGetBitmap(mappage);
    Assert(ISSET(freep, bitmapbit));
    CLRBIT(freep, bitmapbit);
    _hash_wrtbuf(rel, mapbuf);

    /* Get write-lock on metapage to update firstfree */
    _hash_chgbufaccess(rel, metabuf, HASH_NOLOCK, HASH_WRITE);

    /* if this is now the first free page, update hashm_firstfree */
    if (ovflbitno < metap->hashm_firstfree)
    {
        metap->hashm_firstfree = ovflbitno;
        _hash_wrtbuf(rel, metabuf);
    }
    else
    {
        /* no need to change metapage */
        _hash_relbuf(rel, metabuf);
    }

    return nextblkno;
}
Exemple #3
0
/*
 *	_hash_metapinit() -- Initialize the metadata page of a hash index,
 *				the two buckets that we begin with and the initial
 *				bitmap page.
 *
 * We are fairly cavalier about locking here, since we know that no one else
 * could be accessing this index.  In particular the rule about not holding
 * multiple buffer locks is ignored.
 */
void
_hash_metapinit(Relation rel)
{
	HashMetaPage metap;
	HashPageOpaque pageopaque;
	Buffer		metabuf;
	Buffer		buf;
	Page		pg;
	int32		data_width;
	int32		item_width;
	int32		ffactor;
	uint16		i;

	/* safety check */
	if (RelationGetNumberOfBlocks(rel) != 0)
		elog(ERROR, "cannot initialize non-empty hash index \"%s\"",
			 RelationGetRelationName(rel));

	/*
	 * Determine the target fill factor (tuples per bucket) for this index.
	 * The idea is to make the fill factor correspond to pages about 3/4ths
	 * full.  We can compute it exactly if the index datatype is fixed-width,
	 * but for var-width there's some guessing involved.
	 */
	data_width = get_typavgwidth(RelationGetDescr(rel)->attrs[0]->atttypid,
								 RelationGetDescr(rel)->attrs[0]->atttypmod);
	item_width = MAXALIGN(sizeof(HashItemData)) + MAXALIGN(data_width) +
		sizeof(ItemIdData);		/* include the line pointer */
	ffactor = (BLCKSZ * 3 / 4) / item_width;
	/* keep to a sane range */
	if (ffactor < 10)
		ffactor = 10;

	metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_WRITE);
	pg = BufferGetPage(metabuf);
	_hash_pageinit(pg, BufferGetPageSize(metabuf));

	pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
	pageopaque->hasho_prevblkno = InvalidBlockNumber;
	pageopaque->hasho_nextblkno = InvalidBlockNumber;
	pageopaque->hasho_bucket = -1;
	pageopaque->hasho_flag = LH_META_PAGE;
	pageopaque->hasho_filler = HASHO_FILL;

	metap = (HashMetaPage) pg;

	metap->hashm_magic = HASH_MAGIC;
	metap->hashm_version = HASH_VERSION;
	metap->hashm_ntuples = 0;
	metap->hashm_nmaps = 0;
	metap->hashm_ffactor = ffactor;
	metap->hashm_bsize = BufferGetPageSize(metabuf);
	/* find largest bitmap array size that will fit in page size */
	for (i = _hash_log2(metap->hashm_bsize); i > 0; --i)
	{
		if ((1 << i) <= (metap->hashm_bsize -
						 (MAXALIGN(sizeof(PageHeaderData)) +
						  MAXALIGN(sizeof(HashPageOpaqueData)))))
			break;
	}
	Assert(i > 0);
	metap->hashm_bmsize = 1 << i;
	metap->hashm_bmshift = i + BYTE_TO_BIT;
	Assert((1 << BMPG_SHIFT(metap)) == (BMPG_MASK(metap) + 1));

	metap->hashm_procid = index_getprocid(rel, 1, HASHPROC);

	/*
	 * We initialize the index with two buckets, 0 and 1, occupying physical
	 * blocks 1 and 2.  The first freespace bitmap page is in block 3.
	 */
	metap->hashm_maxbucket = metap->hashm_lowmask = 1;	/* nbuckets - 1 */
	metap->hashm_highmask = 3;	/* (nbuckets << 1) - 1 */

	MemSet((char *) metap->hashm_spares, 0, sizeof(metap->hashm_spares));
	MemSet((char *) metap->hashm_mapp, 0, sizeof(metap->hashm_mapp));

	metap->hashm_spares[1] = 1;	/* the first bitmap page is only spare */
	metap->hashm_ovflpoint = 1;
	metap->hashm_firstfree = 0;

	/*
	 * Initialize the first two buckets
	 */
	for (i = 0; i <= 1; i++)
	{
		buf = _hash_getbuf(rel, BUCKET_TO_BLKNO(metap, i), HASH_WRITE);
		pg = BufferGetPage(buf);
		_hash_pageinit(pg, BufferGetPageSize(buf));
		pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
		pageopaque->hasho_prevblkno = InvalidBlockNumber;
		pageopaque->hasho_nextblkno = InvalidBlockNumber;
		pageopaque->hasho_bucket = i;
		pageopaque->hasho_flag = LH_BUCKET_PAGE;
		pageopaque->hasho_filler = HASHO_FILL;
		_hash_wrtbuf(rel, buf);
	}

	/*
	 * Initialize first bitmap page.  Can't do this until we
	 * create the first two buckets, else smgr will complain.
	 */
	_hash_initbitmap(rel, metap, 3);

	/* all done */
	_hash_wrtbuf(rel, metabuf);
}
Exemple #4
0
/* ------------------------------------------------
 * hash_bitmap_info()
 *
 * Get bitmap information for a particular overflow page
 *
 * Usage: SELECT * FROM hash_bitmap_info('con_hash_index'::regclass, 5);
 * ------------------------------------------------
 */
Datum
hash_bitmap_info(PG_FUNCTION_ARGS)
{
	Oid			indexRelid = PG_GETARG_OID(0);
	uint64		ovflblkno = PG_GETARG_INT64(1);
	HashMetaPage metap;
	Buffer		metabuf,
				mapbuf;
	BlockNumber bitmapblkno;
	Page		mappage;
	bool		bit = false;
	TupleDesc	tupleDesc;
	Relation	indexRel;
	uint32		ovflbitno;
	int32		bitmappage,
				bitmapbit;
	HeapTuple	tuple;
	int			i,
				j;
	Datum		values[3];
	bool		nulls[3];
	uint32	   *freep;

	if (!superuser())
		ereport(ERROR,
				(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
				 (errmsg("must be superuser to use raw page functions"))));

	indexRel = index_open(indexRelid, AccessShareLock);

	if (!IS_HASH(indexRel))
		elog(ERROR, "relation \"%s\" is not a hash index",
			 RelationGetRelationName(indexRel));

	if (RELATION_IS_OTHER_TEMP(indexRel))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("cannot access temporary tables of other sessions")));

	if (ovflblkno >= RelationGetNumberOfBlocks(indexRel))
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("block number " UINT64_FORMAT " is out of range for relation \"%s\"",
						ovflblkno, RelationGetRelationName(indexRel))));

	/* Read the metapage so we can determine which bitmap page to use */
	metabuf = _hash_getbuf(indexRel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
	metap = HashPageGetMeta(BufferGetPage(metabuf));

	/*
	 * Reject attempt to read the bit for a metapage or bitmap page; this is
	 * only meaningful for overflow pages.
	 */
	if (ovflblkno == 0)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("invalid overflow block number %u",
						(BlockNumber) ovflblkno)));
	for (i = 0; i < metap->hashm_nmaps; i++)
		if (metap->hashm_mapp[i] == ovflblkno)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid overflow block number %u",
							(BlockNumber) ovflblkno)));

	/*
	 * Identify overflow bit number.  This will error out for primary bucket
	 * pages, and we've already rejected the metapage and bitmap pages above.
	 */
	ovflbitno = _hash_ovflblkno_to_bitno(metap, (BlockNumber) ovflblkno);

	bitmappage = ovflbitno >> BMPG_SHIFT(metap);
	bitmapbit = ovflbitno & BMPG_MASK(metap);

	if (bitmappage >= metap->hashm_nmaps)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("invalid overflow block number %u",
						(BlockNumber) ovflblkno)));

	bitmapblkno = metap->hashm_mapp[bitmappage];

	_hash_relbuf(indexRel, metabuf);

	/* Check the status of bitmap bit for overflow page */
	mapbuf = _hash_getbuf(indexRel, bitmapblkno, HASH_READ, LH_BITMAP_PAGE);
	mappage = BufferGetPage(mapbuf);
	freep = HashPageGetBitmap(mappage);

	bit = ISSET(freep, bitmapbit) != 0;

	_hash_relbuf(indexRel, mapbuf);
	index_close(indexRel, AccessShareLock);

	/* Build a tuple descriptor for our result type */
	if (get_call_result_type(fcinfo, NULL, &tupleDesc) != TYPEFUNC_COMPOSITE)
		elog(ERROR, "return type must be a row type");
	tupleDesc = BlessTupleDesc(tupleDesc);

	MemSet(nulls, 0, sizeof(nulls));

	j = 0;
	values[j++] = Int64GetDatum((int64) bitmapblkno);
	values[j++] = Int32GetDatum(bitmapbit);
	values[j++] = BoolGetDatum(bit);

	tuple = heap_form_tuple(tupleDesc, values, nulls);

	PG_RETURN_DATUM(HeapTupleGetDatum(tuple));
}
Exemple #5
0
/*
 *	_hash_metapinit() -- Initialize the metadata page of a hash index,
 *				the initial buckets, and the initial bitmap page.
 *
 * The initial number of buckets is dependent on num_tuples, an estimate
 * of the number of tuples to be loaded into the index initially.  The
 * chosen number of buckets is returned.
 *
 * We are fairly cavalier about locking here, since we know that no one else
 * could be accessing this index.  In particular the rule about not holding
 * multiple buffer locks is ignored.
 */
uint32
_hash_metapinit(Relation rel, double num_tuples, ForkNumber forkNum)
{
    HashMetaPage metap;
    HashPageOpaque pageopaque;
    Buffer		metabuf;
    Buffer		buf;
    Page		pg;
    int32		data_width;
    int32		item_width;
    int32		ffactor;
    double		dnumbuckets;
    uint32		num_buckets;
    uint32		log2_num_buckets;
    uint32		i;

    /* safety check */
    if (RelationGetNumberOfBlocksInFork(rel, forkNum) != 0)
        elog(ERROR, "cannot initialize non-empty hash index \"%s\"",
             RelationGetRelationName(rel));

    /*
     * Determine the target fill factor (in tuples per bucket) for this index.
     * The idea is to make the fill factor correspond to pages about as full
     * as the user-settable fillfactor parameter says.	We can compute it
     * exactly since the index datatype (i.e. uint32 hash key) is fixed-width.
     */
    data_width = sizeof(uint32);
    item_width = MAXALIGN(sizeof(IndexTupleData)) + MAXALIGN(data_width) +
                 sizeof(ItemIdData);		/* include the line pointer */
    ffactor = RelationGetTargetPageUsage(rel, HASH_DEFAULT_FILLFACTOR) / item_width;
    /* keep to a sane range */
    if (ffactor < 10)
        ffactor = 10;

    /*
     * Choose the number of initial bucket pages to match the fill factor
     * given the estimated number of tuples.  We round up the result to the
     * next power of 2, however, and always force at least 2 bucket pages. The
     * upper limit is determined by considerations explained in
     * _hash_expandtable().
     */
    dnumbuckets = num_tuples / ffactor;
    if (dnumbuckets <= 2.0)
        num_buckets = 2;
    else if (dnumbuckets >= (double) 0x40000000)
        num_buckets = 0x40000000;
    else
        num_buckets = ((uint32) 1) << _hash_log2((uint32) dnumbuckets);

    log2_num_buckets = _hash_log2(num_buckets);
    Assert(num_buckets == (((uint32) 1) << log2_num_buckets));
    Assert(log2_num_buckets < HASH_MAX_SPLITPOINTS);

    /*
     * We initialize the metapage, the first N bucket pages, and the first
     * bitmap page in sequence, using _hash_getnewbuf to cause smgrextend()
     * calls to occur.	This ensures that the smgr level has the right idea of
     * the physical index length.
     */
    metabuf = _hash_getnewbuf(rel, HASH_METAPAGE, forkNum);
    pg = BufferGetPage(metabuf);

    pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
    pageopaque->hasho_prevblkno = InvalidBlockNumber;
    pageopaque->hasho_nextblkno = InvalidBlockNumber;
    pageopaque->hasho_bucket = -1;
    pageopaque->hasho_flag = LH_META_PAGE;
    pageopaque->hasho_page_id = HASHO_PAGE_ID;

    metap = HashPageGetMeta(pg);

    metap->hashm_magic = HASH_MAGIC;
    metap->hashm_version = HASH_VERSION;
    metap->hashm_ntuples = 0;
    metap->hashm_nmaps = 0;
    metap->hashm_ffactor = ffactor;
    metap->hashm_bsize = HashGetMaxBitmapSize(pg);
    /* find largest bitmap array size that will fit in page size */
    for (i = _hash_log2(metap->hashm_bsize); i > 0; --i)
    {
        if ((1 << i) <= metap->hashm_bsize)
            break;
    }
    Assert(i > 0);
    metap->hashm_bmsize = 1 << i;
    metap->hashm_bmshift = i + BYTE_TO_BIT;
    Assert((1 << BMPG_SHIFT(metap)) == (BMPG_MASK(metap) + 1));

    /*
     * Label the index with its primary hash support function's OID.  This is
     * pretty useless for normal operation (in fact, hashm_procid is not used
     * anywhere), but it might be handy for forensic purposes so we keep it.
     */
    metap->hashm_procid = index_getprocid(rel, 1, HASHPROC);

    /*
     * We initialize the index with N buckets, 0 .. N-1, occupying physical
     * blocks 1 to N.  The first freespace bitmap page is in block N+1. Since
     * N is a power of 2, we can set the masks this way:
     */
    metap->hashm_maxbucket = metap->hashm_lowmask = num_buckets - 1;
    metap->hashm_highmask = (num_buckets << 1) - 1;

    MemSet(metap->hashm_spares, 0, sizeof(metap->hashm_spares));
    MemSet(metap->hashm_mapp, 0, sizeof(metap->hashm_mapp));

    /* Set up mapping for one spare page after the initial splitpoints */
    metap->hashm_spares[log2_num_buckets] = 1;
    metap->hashm_ovflpoint = log2_num_buckets;
    metap->hashm_firstfree = 0;

    /*
     * Release buffer lock on the metapage while we initialize buckets.
     * Otherwise, we'll be in interrupt holdoff and the CHECK_FOR_INTERRUPTS
     * won't accomplish anything.  It's a bad idea to hold buffer locks for
     * long intervals in any case, since that can block the bgwriter.
     */
    _hash_chgbufaccess(rel, metabuf, HASH_WRITE, HASH_NOLOCK);

    /*
     * Initialize the first N buckets
     */
    for (i = 0; i < num_buckets; i++)
    {
        /* Allow interrupts, in case N is huge */
        CHECK_FOR_INTERRUPTS();

        buf = _hash_getnewbuf(rel, BUCKET_TO_BLKNO(metap, i), forkNum);
        pg = BufferGetPage(buf);
        pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
        pageopaque->hasho_prevblkno = InvalidBlockNumber;
        pageopaque->hasho_nextblkno = InvalidBlockNumber;
        pageopaque->hasho_bucket = i;
        pageopaque->hasho_flag = LH_BUCKET_PAGE;
        pageopaque->hasho_page_id = HASHO_PAGE_ID;
        _hash_wrtbuf(rel, buf);
    }

    /* Now reacquire buffer lock on metapage */
    _hash_chgbufaccess(rel, metabuf, HASH_NOLOCK, HASH_WRITE);

    /*
     * Initialize first bitmap page
     */
    _hash_initbitmap(rel, metap, num_buckets + 1, forkNum);

    /* all done */
    _hash_wrtbuf(rel, metabuf);

    return num_buckets;
}
Exemple #6
0
/*
 *	_hash_freeovflpage() -
 *
 *	Remove this overflow page from its bucket's chain, and mark the page as
 *	free.  On entry, ovflbuf is write-locked; it is released before exiting.
 *
 *	Add the tuples (itups) to wbuf in this function.  We could do that in the
 *	caller as well, but the advantage of doing it here is we can easily write
 *	the WAL for XLOG_HASH_SQUEEZE_PAGE operation.  Addition of tuples and
 *	removal of overflow page has to done as an atomic operation, otherwise
 *	during replay on standby users might find duplicate records.
 *
 *	Since this function is invoked in VACUUM, we provide an access strategy
 *	parameter that controls fetches of the bucket pages.
 *
 *	Returns the block number of the page that followed the given page
 *	in the bucket, or InvalidBlockNumber if no following page.
 *
 *	NB: caller must not hold lock on metapage, nor on page, that's next to
 *	ovflbuf in the bucket chain.  We don't acquire the lock on page that's
 *	prior to ovflbuf in chain if it is same as wbuf because the caller already
 *	has a lock on same.
 */
BlockNumber
_hash_freeovflpage(Relation rel, Buffer bucketbuf, Buffer ovflbuf,
				   Buffer wbuf, IndexTuple *itups, OffsetNumber *itup_offsets,
				   Size *tups_size, uint16 nitups,
				   BufferAccessStrategy bstrategy)
{
	HashMetaPage metap;
	Buffer		metabuf;
	Buffer		mapbuf;
	BlockNumber ovflblkno;
	BlockNumber prevblkno;
	BlockNumber blkno;
	BlockNumber nextblkno;
	BlockNumber writeblkno;
	HashPageOpaque ovflopaque;
	Page		ovflpage;
	Page		mappage;
	uint32	   *freep;
	uint32		ovflbitno;
	int32		bitmappage,
				bitmapbit;
	Bucket		bucket PG_USED_FOR_ASSERTS_ONLY;
	Buffer		prevbuf = InvalidBuffer;
	Buffer		nextbuf = InvalidBuffer;
	bool		update_metap = false;

	/* Get information from the doomed page */
	_hash_checkpage(rel, ovflbuf, LH_OVERFLOW_PAGE);
	ovflblkno = BufferGetBlockNumber(ovflbuf);
	ovflpage = BufferGetPage(ovflbuf);
	ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
	nextblkno = ovflopaque->hasho_nextblkno;
	prevblkno = ovflopaque->hasho_prevblkno;
	writeblkno = BufferGetBlockNumber(wbuf);
	bucket = ovflopaque->hasho_bucket;

	/*
	 * Fix up the bucket chain.  this is a doubly-linked list, so we must fix
	 * up the bucket chain members behind and ahead of the overflow page being
	 * deleted.  Concurrency issues are avoided by using lock chaining as
	 * described atop hashbucketcleanup.
	 */
	if (BlockNumberIsValid(prevblkno))
	{
		if (prevblkno == writeblkno)
			prevbuf = wbuf;
		else
			prevbuf = _hash_getbuf_with_strategy(rel,
												 prevblkno,
												 HASH_WRITE,
												 LH_BUCKET_PAGE | LH_OVERFLOW_PAGE,
												 bstrategy);
	}
	if (BlockNumberIsValid(nextblkno))
		nextbuf = _hash_getbuf_with_strategy(rel,
											 nextblkno,
											 HASH_WRITE,
											 LH_OVERFLOW_PAGE,
											 bstrategy);

	/* Note: bstrategy is intentionally not used for metapage and bitmap */

	/* Read the metapage so we can determine which bitmap page to use */
	metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
	metap = HashPageGetMeta(BufferGetPage(metabuf));

	/* Identify which bit to set */
	ovflbitno = _hash_ovflblkno_to_bitno(metap, ovflblkno);

	bitmappage = ovflbitno >> BMPG_SHIFT(metap);
	bitmapbit = ovflbitno & BMPG_MASK(metap);

	if (bitmappage >= metap->hashm_nmaps)
		elog(ERROR, "invalid overflow bit number %u", ovflbitno);
	blkno = metap->hashm_mapp[bitmappage];

	/* Release metapage lock while we access the bitmap page */
	LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);

	/* read the bitmap page to clear the bitmap bit */
	mapbuf = _hash_getbuf(rel, blkno, HASH_WRITE, LH_BITMAP_PAGE);
	mappage = BufferGetPage(mapbuf);
	freep = HashPageGetBitmap(mappage);
	Assert(ISSET(freep, bitmapbit));

	/* Get write-lock on metapage to update firstfree */
	LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);

	/* This operation needs to log multiple tuples, prepare WAL for that */
	if (RelationNeedsWAL(rel))
		XLogEnsureRecordSpace(HASH_XLOG_FREE_OVFL_BUFS, 4 + nitups);

	START_CRIT_SECTION();

	/*
	 * we have to insert tuples on the "write" page, being careful to preserve
	 * hashkey ordering.  (If we insert many tuples into the same "write" page
	 * it would be worth qsort'ing them).
	 */
	if (nitups > 0)
	{
		_hash_pgaddmultitup(rel, wbuf, itups, itup_offsets, nitups);
		MarkBufferDirty(wbuf);
	}

	/*
	 * Reinitialize the freed overflow page.  Just zeroing the page won't
	 * work, because WAL replay routines expect pages to be initialized. See
	 * explanation of RBM_NORMAL mode atop XLogReadBufferExtended.  We are
	 * careful to make the special space valid here so that tools like
	 * pageinspect won't get confused.
	 */
	_hash_pageinit(ovflpage, BufferGetPageSize(ovflbuf));

	ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);

	ovflopaque->hasho_prevblkno = InvalidBlockNumber;
	ovflopaque->hasho_nextblkno = InvalidBlockNumber;
	ovflopaque->hasho_bucket = -1;
	ovflopaque->hasho_flag = LH_UNUSED_PAGE;
	ovflopaque->hasho_page_id = HASHO_PAGE_ID;

	MarkBufferDirty(ovflbuf);

	if (BufferIsValid(prevbuf))
	{
		Page		prevpage = BufferGetPage(prevbuf);
		HashPageOpaque prevopaque = (HashPageOpaque) PageGetSpecialPointer(prevpage);

		Assert(prevopaque->hasho_bucket == bucket);
		prevopaque->hasho_nextblkno = nextblkno;
		MarkBufferDirty(prevbuf);
	}
	if (BufferIsValid(nextbuf))
	{
		Page		nextpage = BufferGetPage(nextbuf);
		HashPageOpaque nextopaque = (HashPageOpaque) PageGetSpecialPointer(nextpage);

		Assert(nextopaque->hasho_bucket == bucket);
		nextopaque->hasho_prevblkno = prevblkno;
		MarkBufferDirty(nextbuf);
	}

	/* Clear the bitmap bit to indicate that this overflow page is free */
	CLRBIT(freep, bitmapbit);
	MarkBufferDirty(mapbuf);

	/* if this is now the first free page, update hashm_firstfree */
	if (ovflbitno < metap->hashm_firstfree)
	{
		metap->hashm_firstfree = ovflbitno;
		update_metap = true;
		MarkBufferDirty(metabuf);
	}

	/* XLOG stuff */
	if (RelationNeedsWAL(rel))
	{
		xl_hash_squeeze_page xlrec;
		XLogRecPtr	recptr;
		int			i;

		xlrec.prevblkno = prevblkno;
		xlrec.nextblkno = nextblkno;
		xlrec.ntups = nitups;
		xlrec.is_prim_bucket_same_wrt = (wbuf == bucketbuf);
		xlrec.is_prev_bucket_same_wrt = (wbuf == prevbuf);

		XLogBeginInsert();
		XLogRegisterData((char *) &xlrec, SizeOfHashSqueezePage);

		/*
		 * bucket buffer needs to be registered to ensure that we can acquire
		 * a cleanup lock on it during replay.
		 */
		if (!xlrec.is_prim_bucket_same_wrt)
			XLogRegisterBuffer(0, bucketbuf, REGBUF_STANDARD | REGBUF_NO_IMAGE);

		XLogRegisterBuffer(1, wbuf, REGBUF_STANDARD);
		if (xlrec.ntups > 0)
		{
			XLogRegisterBufData(1, (char *) itup_offsets,
								nitups * sizeof(OffsetNumber));
			for (i = 0; i < nitups; i++)
				XLogRegisterBufData(1, (char *) itups[i], tups_size[i]);
		}

		XLogRegisterBuffer(2, ovflbuf, REGBUF_STANDARD);

		/*
		 * If prevpage and the writepage (block in which we are moving tuples
		 * from overflow) are same, then no need to separately register
		 * prevpage.  During replay, we can directly update the nextblock in
		 * writepage.
		 */
		if (BufferIsValid(prevbuf) && !xlrec.is_prev_bucket_same_wrt)
			XLogRegisterBuffer(3, prevbuf, REGBUF_STANDARD);

		if (BufferIsValid(nextbuf))
			XLogRegisterBuffer(4, nextbuf, REGBUF_STANDARD);

		XLogRegisterBuffer(5, mapbuf, REGBUF_STANDARD);
		XLogRegisterBufData(5, (char *) &bitmapbit, sizeof(uint32));

		if (update_metap)
		{
			XLogRegisterBuffer(6, metabuf, REGBUF_STANDARD);
			XLogRegisterBufData(6, (char *) &metap->hashm_firstfree, sizeof(uint32));
		}

		recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_SQUEEZE_PAGE);

		PageSetLSN(BufferGetPage(wbuf), recptr);
		PageSetLSN(BufferGetPage(ovflbuf), recptr);

		if (BufferIsValid(prevbuf) && !xlrec.is_prev_bucket_same_wrt)
			PageSetLSN(BufferGetPage(prevbuf), recptr);
		if (BufferIsValid(nextbuf))
			PageSetLSN(BufferGetPage(nextbuf), recptr);

		PageSetLSN(BufferGetPage(mapbuf), recptr);

		if (update_metap)
			PageSetLSN(BufferGetPage(metabuf), recptr);
	}

	END_CRIT_SECTION();

	/* release previous bucket if it is not same as write bucket */
	if (BufferIsValid(prevbuf) && prevblkno != writeblkno)
		_hash_relbuf(rel, prevbuf);

	if (BufferIsValid(ovflbuf))
		_hash_relbuf(rel, ovflbuf);

	if (BufferIsValid(nextbuf))
		_hash_relbuf(rel, nextbuf);

	_hash_relbuf(rel, mapbuf);
	_hash_relbuf(rel, metabuf);

	return nextblkno;
}
Exemple #7
0
/*
 *	_hash_addovflpage
 *
 *	Add an overflow page to the bucket whose last page is pointed to by 'buf'.
 *
 *	On entry, the caller must hold a pin but no lock on 'buf'.  The pin is
 *	dropped before exiting (we assume the caller is not interested in 'buf'
 *	anymore) if not asked to retain.  The pin will be retained only for the
 *	primary bucket.  The returned overflow page will be pinned and
 *	write-locked; it is guaranteed to be empty.
 *
 *	The caller must hold a pin, but no lock, on the metapage buffer.
 *	That buffer is returned in the same state.
 *
 * NB: since this could be executed concurrently by multiple processes,
 * one should not assume that the returned overflow page will be the
 * immediate successor of the originally passed 'buf'.  Additional overflow
 * pages might have been added to the bucket chain in between.
 */
Buffer
_hash_addovflpage(Relation rel, Buffer metabuf, Buffer buf, bool retain_pin)
{
	Buffer		ovflbuf;
	Page		page;
	Page		ovflpage;
	HashPageOpaque pageopaque;
	HashPageOpaque ovflopaque;
	HashMetaPage metap;
	Buffer		mapbuf = InvalidBuffer;
	Buffer		newmapbuf = InvalidBuffer;
	BlockNumber blkno;
	uint32		orig_firstfree;
	uint32		splitnum;
	uint32	   *freep = NULL;
	uint32		max_ovflpg;
	uint32		bit;
	uint32		bitmap_page_bit;
	uint32		first_page;
	uint32		last_bit;
	uint32		last_page;
	uint32		i,
				j;
	bool		page_found = false;

	/*
	 * Write-lock the tail page.  Here, we need to maintain locking order such
	 * that, first acquire the lock on tail page of bucket, then on meta page
	 * to find and lock the bitmap page and if it is found, then lock on meta
	 * page is released, then finally acquire the lock on new overflow buffer.
	 * We need this locking order to avoid deadlock with backends that are
	 * doing inserts.
	 *
	 * Note: We could have avoided locking many buffers here if we made two
	 * WAL records for acquiring an overflow page (one to allocate an overflow
	 * page and another to add it to overflow bucket chain).  However, doing
	 * so can leak an overflow page, if the system crashes after allocation.
	 * Needless to say, it is better to have a single record from a
	 * performance point of view as well.
	 */
	LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);

	/* probably redundant... */
	_hash_checkpage(rel, buf, LH_BUCKET_PAGE | LH_OVERFLOW_PAGE);

	/* loop to find current tail page, in case someone else inserted too */
	for (;;)
	{
		BlockNumber nextblkno;

		page = BufferGetPage(buf);
		pageopaque = (HashPageOpaque) PageGetSpecialPointer(page);
		nextblkno = pageopaque->hasho_nextblkno;

		if (!BlockNumberIsValid(nextblkno))
			break;

		/* we assume we do not need to write the unmodified page */
		if (retain_pin)
		{
			/* pin will be retained only for the primary bucket page */
			Assert((pageopaque->hasho_flag & LH_PAGE_TYPE) == LH_BUCKET_PAGE);
			LockBuffer(buf, BUFFER_LOCK_UNLOCK);
		}
		else
			_hash_relbuf(rel, buf);

		retain_pin = false;

		buf = _hash_getbuf(rel, nextblkno, HASH_WRITE, LH_OVERFLOW_PAGE);
	}

	/* Get exclusive lock on the meta page */
	LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);

	_hash_checkpage(rel, metabuf, LH_META_PAGE);
	metap = HashPageGetMeta(BufferGetPage(metabuf));

	/* start search at hashm_firstfree */
	orig_firstfree = metap->hashm_firstfree;
	first_page = orig_firstfree >> BMPG_SHIFT(metap);
	bit = orig_firstfree & BMPG_MASK(metap);
	i = first_page;
	j = bit / BITS_PER_MAP;
	bit &= ~(BITS_PER_MAP - 1);

	/* outer loop iterates once per bitmap page */
	for (;;)
	{
		BlockNumber mapblkno;
		Page		mappage;
		uint32		last_inpage;

		/* want to end search with the last existing overflow page */
		splitnum = metap->hashm_ovflpoint;
		max_ovflpg = metap->hashm_spares[splitnum] - 1;
		last_page = max_ovflpg >> BMPG_SHIFT(metap);
		last_bit = max_ovflpg & BMPG_MASK(metap);

		if (i > last_page)
			break;

		Assert(i < metap->hashm_nmaps);
		mapblkno = metap->hashm_mapp[i];

		if (i == last_page)
			last_inpage = last_bit;
		else
			last_inpage = BMPGSZ_BIT(metap) - 1;

		/* Release exclusive lock on metapage while reading bitmap page */
		LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);

		mapbuf = _hash_getbuf(rel, mapblkno, HASH_WRITE, LH_BITMAP_PAGE);
		mappage = BufferGetPage(mapbuf);
		freep = HashPageGetBitmap(mappage);

		for (; bit <= last_inpage; j++, bit += BITS_PER_MAP)
		{
			if (freep[j] != ALL_SET)
			{
				page_found = true;

				/* Reacquire exclusive lock on the meta page */
				LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);

				/* convert bit to bit number within page */
				bit += _hash_firstfreebit(freep[j]);
				bitmap_page_bit = bit;

				/* convert bit to absolute bit number */
				bit += (i << BMPG_SHIFT(metap));
				/* Calculate address of the recycled overflow page */
				blkno = bitno_to_blkno(metap, bit);

				/* Fetch and init the recycled page */
				ovflbuf = _hash_getinitbuf(rel, blkno);

				goto found;
			}
		}

		/* No free space here, try to advance to next map page */
		_hash_relbuf(rel, mapbuf);
		mapbuf = InvalidBuffer;
		i++;
		j = 0;					/* scan from start of next map page */
		bit = 0;

		/* Reacquire exclusive lock on the meta page */
		LockBuffer(metabuf, BUFFER_LOCK_EXCLUSIVE);
	}

	/*
	 * No free pages --- have to extend the relation to add an overflow page.
	 * First, check to see if we have to add a new bitmap page too.
	 */
	if (last_bit == (uint32) (BMPGSZ_BIT(metap) - 1))
	{
		/*
		 * We create the new bitmap page with all pages marked "in use".
		 * Actually two pages in the new bitmap's range will exist
		 * immediately: the bitmap page itself, and the following page which
		 * is the one we return to the caller.  Both of these are correctly
		 * marked "in use".  Subsequent pages do not exist yet, but it is
		 * convenient to pre-mark them as "in use" too.
		 */
		bit = metap->hashm_spares[splitnum];

		/* metapage already has a write lock */
		if (metap->hashm_nmaps >= HASH_MAX_BITMAPS)
			ereport(ERROR,
					(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
					 errmsg("out of overflow pages in hash index \"%s\"",
							RelationGetRelationName(rel))));

		newmapbuf = _hash_getnewbuf(rel, bitno_to_blkno(metap, bit), MAIN_FORKNUM);
	}
	else
	{
		/*
		 * Nothing to do here; since the page will be past the last used page,
		 * we know its bitmap bit was preinitialized to "in use".
		 */
	}

	/* Calculate address of the new overflow page */
	bit = BufferIsValid(newmapbuf) ?
		metap->hashm_spares[splitnum] + 1 : metap->hashm_spares[splitnum];
	blkno = bitno_to_blkno(metap, bit);

	/*
	 * Fetch the page with _hash_getnewbuf to ensure smgr's idea of the
	 * relation length stays in sync with ours.  XXX It's annoying to do this
	 * with metapage write lock held; would be better to use a lock that
	 * doesn't block incoming searches.
	 *
	 * It is okay to hold two buffer locks here (one on tail page of bucket
	 * and other on new overflow page) since there cannot be anyone else
	 * contending for access to ovflbuf.
	 */
	ovflbuf = _hash_getnewbuf(rel, blkno, MAIN_FORKNUM);

found:

	/*
	 * Do the update.  No ereport(ERROR) until changes are logged. We want to
	 * log the changes for bitmap page and overflow page together to avoid
	 * loss of pages in case the new page is added.
	 */
	START_CRIT_SECTION();

	if (page_found)
	{
		Assert(BufferIsValid(mapbuf));

		/* mark page "in use" in the bitmap */
		SETBIT(freep, bitmap_page_bit);
		MarkBufferDirty(mapbuf);
	}
	else
	{
		/* update the count to indicate new overflow page is added */
		metap->hashm_spares[splitnum]++;

		if (BufferIsValid(newmapbuf))
		{
			_hash_initbitmapbuffer(newmapbuf, metap->hashm_bmsize, false);
			MarkBufferDirty(newmapbuf);

			/* add the new bitmap page to the metapage's list of bitmaps */
			metap->hashm_mapp[metap->hashm_nmaps] = BufferGetBlockNumber(newmapbuf);
			metap->hashm_nmaps++;
			metap->hashm_spares[splitnum]++;
		}

		MarkBufferDirty(metabuf);

		/*
		 * for new overflow page, we don't need to explicitly set the bit in
		 * bitmap page, as by default that will be set to "in use".
		 */
	}

	/*
	 * Adjust hashm_firstfree to avoid redundant searches.  But don't risk
	 * changing it if someone moved it while we were searching bitmap pages.
	 */
	if (metap->hashm_firstfree == orig_firstfree)
	{
		metap->hashm_firstfree = bit + 1;
		MarkBufferDirty(metabuf);
	}

	/* initialize new overflow page */
	ovflpage = BufferGetPage(ovflbuf);
	ovflopaque = (HashPageOpaque) PageGetSpecialPointer(ovflpage);
	ovflopaque->hasho_prevblkno = BufferGetBlockNumber(buf);
	ovflopaque->hasho_nextblkno = InvalidBlockNumber;
	ovflopaque->hasho_bucket = pageopaque->hasho_bucket;
	ovflopaque->hasho_flag = LH_OVERFLOW_PAGE;
	ovflopaque->hasho_page_id = HASHO_PAGE_ID;

	MarkBufferDirty(ovflbuf);

	/* logically chain overflow page to previous page */
	pageopaque->hasho_nextblkno = BufferGetBlockNumber(ovflbuf);

	MarkBufferDirty(buf);

	/* XLOG stuff */
	if (RelationNeedsWAL(rel))
	{
		XLogRecPtr	recptr;
		xl_hash_add_ovfl_page xlrec;

		xlrec.bmpage_found = page_found;
		xlrec.bmsize = metap->hashm_bmsize;

		XLogBeginInsert();
		XLogRegisterData((char *) &xlrec, SizeOfHashAddOvflPage);

		XLogRegisterBuffer(0, ovflbuf, REGBUF_WILL_INIT);
		XLogRegisterBufData(0, (char *) &pageopaque->hasho_bucket, sizeof(Bucket));

		XLogRegisterBuffer(1, buf, REGBUF_STANDARD);

		if (BufferIsValid(mapbuf))
		{
			XLogRegisterBuffer(2, mapbuf, REGBUF_STANDARD);
			XLogRegisterBufData(2, (char *) &bitmap_page_bit, sizeof(uint32));
		}

		if (BufferIsValid(newmapbuf))
			XLogRegisterBuffer(3, newmapbuf, REGBUF_WILL_INIT);

		XLogRegisterBuffer(4, metabuf, REGBUF_STANDARD);
		XLogRegisterBufData(4, (char *) &metap->hashm_firstfree, sizeof(uint32));

		recptr = XLogInsert(RM_HASH_ID, XLOG_HASH_ADD_OVFL_PAGE);

		PageSetLSN(BufferGetPage(ovflbuf), recptr);
		PageSetLSN(BufferGetPage(buf), recptr);

		if (BufferIsValid(mapbuf))
			PageSetLSN(BufferGetPage(mapbuf), recptr);

		if (BufferIsValid(newmapbuf))
			PageSetLSN(BufferGetPage(newmapbuf), recptr);

		PageSetLSN(BufferGetPage(metabuf), recptr);
	}

	END_CRIT_SECTION();

	if (retain_pin)
		LockBuffer(buf, BUFFER_LOCK_UNLOCK);
	else
		_hash_relbuf(rel, buf);

	if (BufferIsValid(mapbuf))
		_hash_relbuf(rel, mapbuf);

	LockBuffer(metabuf, BUFFER_LOCK_UNLOCK);

	if (BufferIsValid(newmapbuf))
		_hash_relbuf(rel, newmapbuf);

	return ovflbuf;
}
Exemple #8
0
/*
 *	_hash_metapinit() -- Initialize the metadata page of a hash index,
 *				the two buckets that we begin with and the initial
 *				bitmap page.
 *
 * We are fairly cavalier about locking here, since we know that no one else
 * could be accessing this index.  In particular the rule about not holding
 * multiple buffer locks is ignored.
 */
void
_hash_metapinit(Relation rel)
{
	MIRROREDLOCK_BUFMGR_DECLARE;

	HashMetaPage metap;
	HashPageOpaque pageopaque;
	Buffer		metabuf;
	Buffer		buf;
	Page		pg;
	int32		data_width;
	int32		item_width;
	int32		ffactor;
	uint16		i;

	/* safety check */
	if (RelationGetNumberOfBlocks(rel) != 0)
		elog(ERROR, "cannot initialize non-empty hash index \"%s\"",
			 RelationGetRelationName(rel));

	/*
	 * Determine the target fill factor (in tuples per bucket) for this index.
	 * The idea is to make the fill factor correspond to pages about as full
	 * as the user-settable fillfactor parameter says.	We can compute it
	 * exactly if the index datatype is fixed-width, but for var-width there's
	 * some guessing involved.
	 */
	data_width = get_typavgwidth(RelationGetDescr(rel)->attrs[0]->atttypid,
								 RelationGetDescr(rel)->attrs[0]->atttypmod);
	item_width = MAXALIGN(sizeof(IndexTupleData)) + MAXALIGN(data_width) +
		sizeof(ItemIdData);		/* include the line pointer */
	ffactor = RelationGetTargetPageUsage(rel, HASH_DEFAULT_FILLFACTOR) / item_width;
	/* keep to a sane range */
	if (ffactor < 10)
		ffactor = 10;

	/*
	 * We initialize the metapage, the first two bucket pages, and the
	 * first bitmap page in sequence, using _hash_getnewbuf to cause
	 * smgrextend() calls to occur.  This ensures that the smgr level
	 * has the right idea of the physical index length.
	 */
	
	// -------- MirroredLock ----------
	MIRROREDLOCK_BUFMGR_LOCK;
	
	metabuf = _hash_getnewbuf(rel, HASH_METAPAGE, HASH_WRITE);
	pg = BufferGetPage(metabuf);
	_hash_pageinit(pg, BufferGetPageSize(metabuf));

	pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
	pageopaque->hasho_prevblkno = InvalidBlockNumber;
	pageopaque->hasho_nextblkno = InvalidBlockNumber;
	pageopaque->hasho_bucket = -1;
	pageopaque->hasho_flag = LH_META_PAGE;
	pageopaque->hasho_filler = HASHO_FILL;

	metap = (HashMetaPage) pg;

	metap->hashm_magic = HASH_MAGIC;
	metap->hashm_version = HASH_VERSION;
	metap->hashm_ntuples = 0;
	metap->hashm_nmaps = 0;
	metap->hashm_ffactor = ffactor;
	metap->hashm_bsize = BufferGetPageSize(metabuf);
	/* find largest bitmap array size that will fit in page size */
	for (i = _hash_log2(metap->hashm_bsize); i > 0; --i)
	{
		if ((1 << i) <= (metap->hashm_bsize -
						 (MAXALIGN(sizeof(PageHeaderData)) +
						  MAXALIGN(sizeof(HashPageOpaqueData)))))
			break;
	}
	Assert(i > 0);
	metap->hashm_bmsize = 1 << i;
	metap->hashm_bmshift = i + BYTE_TO_BIT;
	Assert((1 << BMPG_SHIFT(metap)) == (BMPG_MASK(metap) + 1));

	metap->hashm_procid = index_getprocid(rel, 1, HASHPROC);

	/*
	 * We initialize the index with two buckets, 0 and 1, occupying physical
	 * blocks 1 and 2.	The first freespace bitmap page is in block 3.
	 */
	metap->hashm_maxbucket = metap->hashm_lowmask = 1;	/* nbuckets - 1 */
	metap->hashm_highmask = 3;	/* (nbuckets << 1) - 1 */

	MemSet(metap->hashm_spares, 0, sizeof(metap->hashm_spares));
	MemSet(metap->hashm_mapp, 0, sizeof(metap->hashm_mapp));

	metap->hashm_spares[1] = 1; /* the first bitmap page is only spare */
	metap->hashm_ovflpoint = 1;
	metap->hashm_firstfree = 0;

	/*
	 * Initialize the first two buckets
	 */
	for (i = 0; i <= 1; i++)
	{
		buf = _hash_getnewbuf(rel, BUCKET_TO_BLKNO(metap, i), HASH_WRITE);
		pg = BufferGetPage(buf);
		_hash_pageinit(pg, BufferGetPageSize(buf));
		pageopaque = (HashPageOpaque) PageGetSpecialPointer(pg);
		pageopaque->hasho_prevblkno = InvalidBlockNumber;
		pageopaque->hasho_nextblkno = InvalidBlockNumber;
		pageopaque->hasho_bucket = i;
		pageopaque->hasho_flag = LH_BUCKET_PAGE;
		pageopaque->hasho_filler = HASHO_FILL;
		_hash_wrtbuf(rel, buf);
	}

	/*
	 * Initialize first bitmap page
	 */
	_hash_initbitmap(rel, metap, 3);

	/* all done */
	_hash_wrtbuf(rel, metabuf);
	
	MIRROREDLOCK_BUFMGR_UNLOCK;
	// -------- MirroredLock ----------
	
}
/*
 *	_hash_init_metabuffer() -- Initialize the metadata page of a hash index.
 */
void
_hash_init_metabuffer(Buffer buf, double num_tuples, RegProcedure procid,
					  uint16 ffactor, bool initpage)
{
	HashMetaPage metap;
	HashPageOpaque pageopaque;
	Page		page;
	double		dnumbuckets;
	uint32		num_buckets;
	uint32		spare_index;
	uint32		i;

	/*
	 * Choose the number of initial bucket pages to match the fill factor
	 * given the estimated number of tuples.  We round up the result to the
	 * total number of buckets which has to be allocated before using its
	 * _hashm_spare element. However always force at least 2 bucket pages. The
	 * upper limit is determined by considerations explained in
	 * _hash_expandtable().
	 */
	dnumbuckets = num_tuples / ffactor;
	if (dnumbuckets <= 2.0)
		num_buckets = 2;
	else if (dnumbuckets >= (double) 0x40000000)
		num_buckets = 0x40000000;
	else
		num_buckets = _hash_get_totalbuckets(_hash_spareindex(dnumbuckets));

	spare_index = _hash_spareindex(num_buckets);
	Assert(spare_index < HASH_MAX_SPLITPOINTS);

	page = BufferGetPage(buf);
	if (initpage)
		_hash_pageinit(page, BufferGetPageSize(buf));

	pageopaque = (HashPageOpaque) PageGetSpecialPointer(page);
	pageopaque->hasho_prevblkno = InvalidBlockNumber;
	pageopaque->hasho_nextblkno = InvalidBlockNumber;
	pageopaque->hasho_bucket = -1;
	pageopaque->hasho_flag = LH_META_PAGE;
	pageopaque->hasho_page_id = HASHO_PAGE_ID;

	metap = HashPageGetMeta(page);

	metap->hashm_magic = HASH_MAGIC;
	metap->hashm_version = HASH_VERSION;
	metap->hashm_ntuples = 0;
	metap->hashm_nmaps = 0;
	metap->hashm_ffactor = ffactor;
	metap->hashm_bsize = HashGetMaxBitmapSize(page);
	/* find largest bitmap array size that will fit in page size */
	for (i = _hash_log2(metap->hashm_bsize); i > 0; --i)
	{
		if ((1 << i) <= metap->hashm_bsize)
			break;
	}
	Assert(i > 0);
	metap->hashm_bmsize = 1 << i;
	metap->hashm_bmshift = i + BYTE_TO_BIT;
	Assert((1 << BMPG_SHIFT(metap)) == (BMPG_MASK(metap) + 1));

	/*
	 * Label the index with its primary hash support function's OID.  This is
	 * pretty useless for normal operation (in fact, hashm_procid is not used
	 * anywhere), but it might be handy for forensic purposes so we keep it.
	 */
	metap->hashm_procid = procid;

	/*
	 * We initialize the index with N buckets, 0 .. N-1, occupying physical
	 * blocks 1 to N.  The first freespace bitmap page is in block N+1.
	 */
	metap->hashm_maxbucket = num_buckets - 1;

	/*
	 * Set highmask as next immediate ((2 ^ x) - 1), which should be
	 * sufficient to cover num_buckets.
	 */
	metap->hashm_highmask = (1 << (_hash_log2(num_buckets + 1))) - 1;
	metap->hashm_lowmask = (metap->hashm_highmask >> 1);

	MemSet(metap->hashm_spares, 0, sizeof(metap->hashm_spares));
	MemSet(metap->hashm_mapp, 0, sizeof(metap->hashm_mapp));

	/* Set up mapping for one spare page after the initial splitpoints */
	metap->hashm_spares[spare_index] = 1;
	metap->hashm_ovflpoint = spare_index;
	metap->hashm_firstfree = 0;

	/*
	 * Set pd_lower just past the end of the metadata.  This is to log full
	 * page image of metapage in xloginsert.c.
	 */
	((PageHeader) page)->pd_lower =
		((char *) metap + sizeof(HashMetaPageData)) - (char *) page;
}