Minor tweaking of index special-space definitions so that the various

index types can be reliably distinguished by examining the special space
on an index page.  Per my earlier proposal, plus the realization that
there's no need for btree's vacuum cycle ID to cycle through every possible
16-bit value.  Restricting its range a little costs nearly nothing and
eliminates the possibility of collisions.
Memo to self: remember to make bitmap indexes play along with this scheme,
assuming that patch ever gets accepted.
This commit is contained in:
Tom Lane 2007-04-09 22:04:08 +00:00
parent 485d9ca96f
commit 56218fbc48
9 changed files with 78 additions and 64 deletions

View File

@ -8,7 +8,7 @@
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/access/gist/gistutil.c,v 1.21 2007/01/05 22:19:22 momjian Exp $
* $PostgreSQL: pgsql/src/backend/access/gist/gistutil.c,v 1.22 2007/04/09 22:03:57 tgl Exp $
*-------------------------------------------------------------------------
*/
#include "postgres.h"
@ -559,10 +559,11 @@ GISTInitBuffer(Buffer b, uint32 f)
PageInit(page, pageSize, sizeof(GISTPageOpaqueData));
opaque = GistPageGetOpaque(page);
opaque->flags = f;
opaque->rightlink = InvalidBlockNumber;
/* page was already zeroed by PageInit, so this is not needed: */
/* memset(&(opaque->nsn), 0, sizeof(GistNSN)); */
opaque->rightlink = InvalidBlockNumber;
opaque->flags = f;
opaque->gist_page_id = GIST_PAGE_ID;
}
/*

View File

@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/access/hash/hashovfl.c,v 1.54 2007/01/05 22:19:22 momjian Exp $
* $PostgreSQL: pgsql/src/backend/access/hash/hashovfl.c,v 1.55 2007/04/09 22:03:57 tgl Exp $
*
* NOTES
* Overflow pages look like ordinary relation pages.
@ -141,7 +141,7 @@ _hash_addovflpage(Relation rel, Buffer metabuf, Buffer buf)
ovflopaque->hasho_nextblkno = InvalidBlockNumber;
ovflopaque->hasho_bucket = pageopaque->hasho_bucket;
ovflopaque->hasho_flag = LH_OVERFLOW_PAGE;
ovflopaque->hasho_filler = HASHO_FILL;
ovflopaque->hasho_page_id = HASHO_PAGE_ID;
MarkBufferDirty(ovflbuf);
@ -529,7 +529,7 @@ _hash_initbitmap(Relation rel, HashMetaPage metap, BlockNumber blkno)
op->hasho_nextblkno = InvalidBlockNumber;
op->hasho_bucket = -1;
op->hasho_flag = LH_BITMAP_PAGE;
op->hasho_filler = HASHO_FILL;
op->hasho_page_id = HASHO_PAGE_ID;
/* set all of the bits to 1 */
freep = HashPageGetBitmap(pg);

View File

@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/access/hash/hashpage.c,v 1.64 2007/01/29 23:22:59 tgl Exp $
* $PostgreSQL: pgsql/src/backend/access/hash/hashpage.c,v 1.65 2007/04/09 22:03:57 tgl Exp $
*
* NOTES
* Postgres hash pages look like ordinary relation pages. The opaque
@ -252,7 +252,7 @@ _hash_metapinit(Relation rel)
pageopaque->hasho_nextblkno = InvalidBlockNumber;
pageopaque->hasho_bucket = -1;
pageopaque->hasho_flag = LH_META_PAGE;
pageopaque->hasho_filler = HASHO_FILL;
pageopaque->hasho_page_id = HASHO_PAGE_ID;
metap = (HashMetaPage) pg;
@ -310,7 +310,7 @@ _hash_metapinit(Relation rel)
pageopaque->hasho_nextblkno = InvalidBlockNumber;
pageopaque->hasho_bucket = i;
pageopaque->hasho_flag = LH_BUCKET_PAGE;
pageopaque->hasho_filler = HASHO_FILL;
pageopaque->hasho_page_id = HASHO_PAGE_ID;
_hash_wrtbuf(rel, buf);
}
@ -654,7 +654,7 @@ _hash_splitbucket(Relation rel,
nopaque->hasho_nextblkno = InvalidBlockNumber;
nopaque->hasho_bucket = nbucket;
nopaque->hasho_flag = LH_BUCKET_PAGE;
nopaque->hasho_filler = HASHO_FILL;
nopaque->hasho_page_id = HASHO_PAGE_ID;
/*
* Partition the tuples in the old bucket between the old bucket and the

View File

@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/access/nbtree/nbtutils.c,v 1.84 2007/04/06 22:33:42 tgl Exp $
* $PostgreSQL: pgsql/src/backend/access/nbtree/nbtutils.c,v 1.85 2007/04/09 22:04:01 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -1263,11 +1263,13 @@ _bt_start_vacuum(Relation rel)
LWLockAcquire(BtreeVacuumLock, LW_EXCLUSIVE);
/* Assign the next cycle ID, being careful to avoid zero */
do
{
result = ++(btvacinfo->cycle_ctr);
} while (result == 0);
/*
* Assign the next cycle ID, being careful to avoid zero as well as
* the reserved high values.
*/
result = ++(btvacinfo->cycle_ctr);
if (result == 0 || result > MAX_BT_CYCLE_ID)
result = btvacinfo->cycle_ctr = 1;
/* Let's just make sure there's no entry already for this index */
for (i = 0; i < btvacinfo->num_vacuums; i++)

View File

@ -3,7 +3,7 @@
* header file for postgres inverted index access method implementation.
*
* Copyright (c) 2006, PostgreSQL Global Development Group
* $PostgreSQL: pgsql/src/include/access/gin.h,v 1.10 2007/01/31 15:09:45 teodor Exp $
* $PostgreSQL: pgsql/src/include/access/gin.h,v 1.11 2007/04/09 22:04:04 tgl Exp $
*--------------------------------------------------------------------------
*/
@ -31,41 +31,27 @@
#define GIN_CONSISTENT_PROC 4
#define GINNProcs 4
typedef XLogRecPtr GinNSN;
/*
* Page opaque data in a inverted index page.
*
* Note: GIN does not include a page ID word as do the other index types.
* This is OK because the opaque data is only 8 bytes and so can be reliably
* distinguished by size. Revisit this if the size ever increases.
*/
typedef struct GinPageOpaqueData
{
uint16 flags;
BlockNumber rightlink; /* next page if any */
OffsetNumber maxoff; /* number entries on GIN_DATA page: number of
* heap ItemPointer on GIN_DATA|GIN_LEAF page
* and number of records on GIN_DATA &
* ~GIN_LEAF page */
BlockNumber rightlink;
uint16 flags; /* see bit definitions below */
} GinPageOpaqueData;
typedef GinPageOpaqueData *GinPageOpaque;
#define GIN_ROOT_BLKNO (0)
typedef struct
{
BlockIdData child_blkno; /* use it instead of BlockNumber to save space
* on page */
ItemPointerData key;
} PostingItem;
#define PostingItemGetBlockNumber(pointer) \
BlockIdGetBlockNumber(&(pointer)->child_blkno)
#define PostingItemSetBlockNumber(pointer, blockNumber) \
BlockIdSet(&((pointer)->child_blkno), (blockNumber))
/*
* Page opaque data in a inverted index page.
*/
#define GIN_DATA (1 << 0)
#define GIN_LEAF (1 << 1)
#define GIN_DELETED (1 << 2)
@ -98,8 +84,21 @@ typedef struct
#define GinItemPointerGetOffsetNumber(pointer) \
((pointer)->ip_posid)
typedef struct
{
BlockIdData child_blkno; /* use it instead of BlockNumber to save space
* on page */
ItemPointerData key;
} PostingItem;
#define PostingItemGetBlockNumber(pointer) \
BlockIdGetBlockNumber(&(pointer)->child_blkno)
#define PostingItemSetBlockNumber(pointer, blockNumber) \
BlockIdSet(&((pointer)->child_blkno), (blockNumber))
/*
* Support work on IndexTuuple on leaf pages
* Support work on IndexTuple on leaf pages
*/
#define GinGetNPosting(itup) GinItemPointerGetOffsetNumber(&(itup)->t_tid)
#define GinSetNPosting(itup,n) ItemPointerSetOffsetNumber(&(itup)->t_tid,(n))

View File

@ -9,7 +9,7 @@
* Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
* $PostgreSQL: pgsql/src/include/access/gist.h,v 1.57 2007/01/05 22:19:50 momjian Exp $
* $PostgreSQL: pgsql/src/include/access/gist.h,v 1.58 2007/04/09 22:04:05 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -56,25 +56,30 @@
/*
* Page opaque data in a GiST index page.
*/
#define F_LEAF (1 << 0)
#define F_DELETED (1 << 1)
#define F_LEAF (1 << 0)
#define F_DELETED (1 << 1)
#define F_TUPLES_DELETED (1 << 2)
typedef XLogRecPtr GistNSN;
typedef struct GISTPageOpaqueData
{
uint32 flags; /* 29 bits are unused for now */
BlockNumber rightlink;
/*
* the only meaning - change this value if page split.
*/
GistNSN nsn;
GistNSN nsn; /* this value must change on page split */
BlockNumber rightlink; /* next page if any */
uint16 flags; /* see bit definitions above */
uint16 gist_page_id; /* for identification of GiST indexes */
} GISTPageOpaqueData;
typedef GISTPageOpaqueData *GISTPageOpaque;
/*
* The page ID is for the convenience of pg_filedump and similar utilities,
* which otherwise would have a hard time telling pages of different index
* types apart. It should be the last 2 bytes on the page. This is more or
* less "free" due to alignment considerations.
*/
#define GIST_PAGE_ID 0xFF81
/*
* This is the Split Vector to be returned by the PickSplit method.
* PickSplit should check spl_(r|l)datum_exists. If it is 'true',

View File

@ -7,7 +7,7 @@
* Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
* $PostgreSQL: pgsql/src/include/access/hash.h,v 1.77 2007/04/02 03:49:40 tgl Exp $
* $PostgreSQL: pgsql/src/include/access/hash.h,v 1.78 2007/04/09 22:04:05 tgl Exp $
*
* NOTES
* modeled after Margo Seltzer's hash implementation for unix.
@ -56,20 +56,18 @@ typedef struct HashPageOpaqueData
BlockNumber hasho_nextblkno; /* next ovfl blkno */
Bucket hasho_bucket; /* bucket number this pg belongs to */
uint16 hasho_flag; /* page type code, see above */
uint16 hasho_filler; /* available for future use */
/*
* We presently set hasho_filler to HASHO_FILL (0x1234); this is for the
* convenience of pg_filedump, which otherwise would have a hard time
* telling HashPageOpaqueData from BTPageOpaqueData. If we ever need that
* space for some other purpose, pg_filedump will have to find another
* way.
*/
uint16 hasho_page_id; /* for identification of hash indexes */
} HashPageOpaqueData;
typedef HashPageOpaqueData *HashPageOpaque;
#define HASHO_FILL 0x1234
/*
* The page ID is for the convenience of pg_filedump and similar utilities,
* which otherwise would have a hard time telling pages of different index
* types apart. It should be the last 2 bytes on the page. This is more or
* less "free" due to alignment considerations.
*/
#define HASHO_PAGE_ID 0xFF80
/*
* HashScanOpaqueData is private state for a hash index scan.

View File

@ -7,7 +7,7 @@
* Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
* $PostgreSQL: pgsql/src/include/access/nbtree.h,v 1.111 2007/02/08 05:05:53 momjian Exp $
* $PostgreSQL: pgsql/src/include/access/nbtree.h,v 1.112 2007/04/09 22:04:08 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -40,8 +40,8 @@ typedef uint16 BTCycleId;
* stored into both halves of the split page. (If VACUUM is not running,
* both pages receive zero cycleids.) This allows VACUUM to detect whether
* a page was split since it started, with a small probability of false match
* if the page was last split some exact multiple of 65536 VACUUMs ago.
* Also, during a split, the BTP_SPLIT_END flag is cleared in the left
* if the page was last split some exact multiple of MAX_BT_CYCLE_ID VACUUMs
* ago. Also, during a split, the BTP_SPLIT_END flag is cleared in the left
* (original) page, and set in the right page, but only if the next page
* to its right has a different cycleid.
*
@ -73,6 +73,15 @@ typedef BTPageOpaqueData *BTPageOpaque;
#define BTP_SPLIT_END (1 << 5) /* rightmost page of split group */
#define BTP_HAS_GARBAGE (1 << 6) /* page has LP_DELETEd tuples */
/*
* The max allowed value of a cycle ID is a bit less than 64K. This is
* for convenience of pg_filedump and similar utilities: we want to use
* the last 2 bytes of special space as an index type indicator, and
* restricting cycle ID lets btree use that space for vacuum cycle IDs
* while still allowing index type to be identified.
*/
#define MAX_BT_CYCLE_ID 0xFF7F
/*
* The Meta page is always the first page in the btree index.

View File

@ -37,7 +37,7 @@
* Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
* $PostgreSQL: pgsql/src/include/catalog/catversion.h,v 1.401 2007/04/06 22:33:43 tgl Exp $
* $PostgreSQL: pgsql/src/include/catalog/catversion.h,v 1.402 2007/04/09 22:04:08 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -53,6 +53,6 @@
*/
/* yyyymmddN */
#define CATALOG_VERSION_NO 200704061
#define CATALOG_VERSION_NO 200704091
#endif