A page image obtained with get_raw_page should be passed as argument:
- test=# SELECT * FROM page_header(get_raw_page('pg_class',0));
- lsn | tli | flags | lower | upper | special | pagesize | version
- ----------+-----+-------+-------+-------+---------+----------+---------
- 0/3C5614 | 1 | 1 | 216 | 256 | 8192 | 8192 | 4
+ regression=# SELECT * FROM page_header(get_raw_page('pg_class',0));
+ lsn | tli | flags | lower | upper | special | pagesize | version | prune_xid
+ -----------+-----+-------+-------+-------+---------+----------+---------+-----------
+ 0/24A1B50 | 1 | 1 | 232 | 368 | 8192 | 8192 | 4 | 0
(1 row)
- The returned columns correspond to the fields in the PageHeaderData-struct,
- see src/include/storage/bufpage.h for more details.
+ The returned columns correspond to the fields in the PageHeaderData struct.
+ See src/include/storage/bufpage.h for details.
heap_page_items
---------------
OUT upper smallint,
OUT special smallint,
OUT pagesize smallint,
- OUT version smallint)
+ OUT version smallint,
+ OUT prune_xid xid)
AS 'MODULE_PATHNAME', 'page_header'
LANGUAGE C STRICT;
* Copyright (c) 2007, PostgreSQL Global Development Group
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/contrib/pageinspect/rawpage.c,v 1.1 2007/05/17 19:11:24 momjian Exp $
+ * $PostgreSQL: pgsql/contrib/pageinspect/rawpage.c,v 1.2 2007/09/21 21:25:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
Datum result;
HeapTuple tuple;
- Datum values[8];
- bool nulls[8];
+ Datum values[9];
+ bool nulls[9];
PageHeader page;
XLogRecPtr lsn;
values[5] = UInt16GetDatum(page->pd_special);
values[6] = UInt16GetDatum(PageGetPageSize(page));
values[7] = UInt16GetDatum(PageGetPageLayoutVersion(page));
+ values[8] = TransactionIdGetDatum(page->pd_prune_xid);
/* Build and return the tuple. */
-<!-- $PostgreSQL: pgsql/doc/src/sgml/storage.sgml,v 1.18 2007/06/03 17:05:53 tgl Exp $ -->
+<!-- $PostgreSQL: pgsql/doc/src/sgml/storage.sgml,v 1.19 2007/09/21 21:25:42 tgl Exp $ -->
<chapter id="storage">
<row>
<entry>PageHeaderData</entry>
- <entry>20 bytes long. Contains general information about the page, including
+ <entry>24 bytes long. Contains general information about the page, including
free space pointers.</entry>
</row>
<para>
- The first 20 bytes of each page consists of a page header
+ The first 24 bytes of each page consists of a page header
(PageHeaderData). Its format is detailed in <xref
linkend="pageheaderdata-table">. The first two fields track the most
recent WAL entry related to this page. Next is a 2-byte field
from the page start to the start
of unallocated space, to the end of unallocated space, and to the start of
the special space.
- The last 2 bytes of the page header,
+ The next 2 bytes of the page header,
<structfield>pd_pagesize_version</structfield>, store both the page size
and a version indicator. Beginning with
<productname>PostgreSQL</productname> 8.3 the version number is 4;
versions, but the layout of heap row headers has.) The page size
is basically only present as a cross-check; there is no support for having
more than one page size in an installation.
+ The last field is a hint that shows whether pruning the page is likely
+ to be profitable: it tracks the oldest un-pruned XMAX on the page.
</para>
<entry>2 bytes</entry>
<entry>Page size and layout version number information</entry>
</row>
+ <row>
+ <entry>pd_prune_xid</entry>
+ <entry>TransactionId</entry>
+ <entry>4 bytes</entry>
+ <entry>Oldest unpruned XMAX on page, or zero if none</entry>
+ </row>
</tbody>
</tgroup>
</table>
-$PostgreSQL: pgsql/src/backend/access/heap/README.HOT,v 1.1 2007/09/20 17:56:30 tgl Exp $
+$PostgreSQL: pgsql/src/backend/access/heap/README.HOT,v 1.2 2007/09/21 21:25:42 tgl Exp $
Heap Only Tuples (HOT)
large enough to accept any extra maintenance burden for.
The currently planned heuristic is to prune and defrag when first accessing
-a page that potentially has prunable tuples (flagged by the PD_PRUNABLE
-page hint bit) and that either has free space less than MAX(fillfactor
+a page that potentially has prunable tuples (as flagged by the pd_prune_xid
+page hint field) and that either has free space less than MAX(fillfactor
target free space, BLCKSZ/10) *or* has recently had an UPDATE fail to
find enough free space to store an updated tuple version. (These rules
are subject to change.)
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/access/heap/heapam.c,v 1.241 2007/09/20 17:56:30 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/access/heap/heapam.c,v 1.242 2007/09/21 21:25:42 tgl Exp $
*
*
* INTERFACE ROUTINES
/*
* If this transaction commits, the tuple will become DEAD sooner or
- * later. Set hint bit that this page is a candidate for pruning. If
- * the transaction finally aborts, the subsequent page pruning will be
- * a no-op and the hint will be cleared.
+ * later. Set flag that this page is a candidate for pruning once our xid
+ * falls below the OldestXmin horizon. If the transaction finally aborts,
+ * the subsequent page pruning will be a no-op and the hint will be
+ * cleared.
*/
- PageSetPrunable((Page) dp);
+ PageSetPrunable(dp, xid);
/* store transaction information of xact deleting the tuple */
tp.t_data->t_infomask &= ~(HEAP_XMAX_COMMITTED |
/*
* If this transaction commits, the old tuple will become DEAD sooner or
- * later. Set hint bit that this page is a candidate for pruning. If
- * the transaction finally aborts, the subsequent page pruning will be
- * a no-op and the hint will be cleared.
+ * later. Set flag that this page is a candidate for pruning once our xid
+ * falls below the OldestXmin horizon. If the transaction finally aborts,
+ * the subsequent page pruning will be a no-op and the hint will be
+ * cleared.
*
* XXX Should we set hint on newbuf as well? If the transaction
* aborts, there would be a prunable tuple in the newbuf; but for now
* we choose not to optimize for aborts. Note that heap_xlog_update
- * must be kept in sync if this changes.
+ * must be kept in sync if this decision changes.
*/
- PageSetPrunable(dp);
+ PageSetPrunable(dp, xid);
if (use_hot_update)
{
HeapTupleHeaderSetCmax(htup, FirstCommandId, false);
/* Mark the page as a candidate for pruning */
- PageSetPrunable(page);
+ PageSetPrunable(page, record->xl_xid);
/* Make sure there is no forward chain link in t_ctid */
htup->t_ctid = xlrec->target.tid;
}
/* Mark the page as a candidate for pruning */
- PageSetPrunable(page);
+ PageSetPrunable(page, record->xl_xid);
/*
* this test is ugly, but necessary to avoid thinking that insert change
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/access/heap/pruneheap.c,v 1.1 2007/09/20 17:56:30 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/access/heap/pruneheap.c,v 1.2 2007/09/21 21:25:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/*
* Let's see if we really need pruning.
*
- * Forget it if page is not hinted to contain something prunable
+ * Forget it if page is not hinted to contain something prunable that's
+ * older than OldestXmin.
*/
- if (!PageIsPrunable(dp))
+ if (!PageIsPrunable(dp, OldestXmin))
return;
/*
/*
* Now that we have buffer lock, get accurate information about the
* page's free space, and recheck the heuristic about whether to prune.
+ * (We needn't recheck PageIsPrunable, since no one else could have
+ * pruned while we hold pin.)
*/
if (PageIsFull(dp) || PageGetHeapFreeSpace((Page) dp) < minfree)
{
START_CRIT_SECTION();
/*
- * Mark the page as clear of prunable tuples. If we find a tuple which
+ * Mark the page as clear of prunable tuples. If we find a tuple which
* may soon become prunable, we shall set the hint again. Also clear
* the "page is full" flag, since there's no point in repeating the
* prune/defrag process until something else happens to the page.
PageSetLSN(BufferGetPage(buffer), recptr);
}
}
+ else
+ {
+ /*
+ * If we didn't prune anything, we have nonetheless updated the
+ * pd_prune_xid field; treat this as a non-WAL-logged hint.
+ */
+ SetBufferCommitInfoNeedsSave(buffer);
+ }
END_CRIT_SECTION();
case HEAPTUPLE_RECENTLY_DEAD:
recent_dead = true;
/*
- * This tuple may soon become DEAD. Re-set the hint bit so
+ * This tuple may soon become DEAD. Update the hint field so
* that the page is reconsidered for pruning in future.
*/
- PageSetPrunable(dp);
+ PageSetPrunable(dp, HeapTupleHeaderGetXmax(htup));
break;
case HEAPTUPLE_DELETE_IN_PROGRESS:
/*
- * This tuple may soon become DEAD. Re-set the hint bit so
+ * This tuple may soon become DEAD. Update the hint field so
* that the page is reconsidered for pruning in future.
*/
- PageSetPrunable(dp);
+ PageSetPrunable(dp, HeapTupleHeaderGetXmax(htup));
break;
case HEAPTUPLE_LIVE:
*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/storage/page/bufpage.c,v 1.74 2007/09/20 17:56:31 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/storage/page/bufpage.c,v 1.75 2007/09/21 21:25:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
/* Make sure all fields of page are zero, as well as unused space */
MemSet(p, 0, pageSize);
- /* p->pd_flags = 0; done by above MemSet */
+ /* p->pd_flags = 0; done by above MemSet */
p->pd_lower = SizeOfPageHeaderData;
p->pd_upper = pageSize - specialSize;
p->pd_special = pageSize - specialSize;
PageSetPageSizeAndVersion(page, pageSize, PG_PAGE_LAYOUT_VERSION);
+ /* p->pd_prune_xid = InvalidTransactionId; done by above MemSet */
}
* 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 1.426 2007/09/20 17:56:32 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/catalog/catversion.h,v 1.427 2007/09/21 21:25:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 200709201
+#define CATALOG_VERSION_NO 200709211
#endif
* Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $PostgreSQL: pgsql/src/include/storage/bufpage.h,v 1.74 2007/09/20 17:56:32 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/storage/bufpage.h,v 1.75 2007/09/21 21:25:42 tgl Exp $
*
*-------------------------------------------------------------------------
*/
*
* EXCEPTIONS:
*
- * obviously, a page is not formatted before it is initialized with by
+ * obviously, a page is not formatted before it is initialized by
* a call to PageInit.
*
* NOTES:
* pd_upper - offset to end of free space.
* pd_special - offset to start of special space.
* pd_pagesize_version - size in bytes and page layout version number.
+ * pd_prune_xid - oldest XID among potentially prunable tuples on page.
*
* The LSN is used by the buffer manager to enforce the basic rule of WAL:
* "thou shalt write xlog before data". A dirty buffer cannot be dumped
* purposes (it is not clear that this is actually necessary, but it seems
* like a good idea).
*
+ * pd_prune_xid is a hint field that helps determine whether pruning will be
+ * useful. It is currently unused in index pages.
+ *
* The page version number and page size are packed together into a single
* uint16 field. This is for historical reasons: before PostgreSQL 7.3,
* there was no concept of a page version number, and doing it this way
LocationIndex pd_upper; /* offset to end of free space */
LocationIndex pd_special; /* offset to start of special space */
uint16 pd_pagesize_version;
+ TransactionId pd_prune_xid; /* oldest prunable XID, or zero if none */
ItemIdData pd_linp[1]; /* beginning of line pointer array */
} PageHeaderData;
* pd_lower. This should be considered a hint rather than the truth, since
* changes to it are not WAL-logged.
*
- * PD_PRUNABLE is set if there are any prunable tuples in the page.
- * This should be considered a hint rather than the truth, since
- * the transaction which generates a prunable tuple may or may not commit.
- * Also there is a lag before a tuple is declared dead.
- *
* PD_PAGE_FULL is set if an UPDATE doesn't find enough free space in the
* page for its new tuple version; this suggests that a prune is needed.
* Again, this is just a hint.
*/
#define PD_HAS_FREE_LINES 0x0001 /* are there any unused line pointers? */
-#define PD_PRUNABLE 0x0002 /* are there any prunable tuples? */
-#define PD_PAGE_FULL 0x0004 /* not enough free space for new tuple? */
+#define PD_PAGE_FULL 0x0002 /* not enough free space for new tuple? */
-#define PD_VALID_FLAG_BITS 0x0007 /* OR of all valid pd_flags bits */
+#define PD_VALID_FLAG_BITS 0x0003 /* OR of all valid pd_flags bits */
/*
* Page layout version number 0 is for pre-7.3 Postgres releases.
* Release 8.0 uses 2; it changed the HeapTupleHeader layout again.
* Release 8.1 uses 3; it redefined HeapTupleHeader infomask bits.
* Release 8.3 uses 4; it changed the HeapTupleHeader layout again, and
- * added the pd_flags field (by stealing some bits from pd_tli).
+ * added the pd_flags field (by stealing some bits from pd_tli),
+ * as well as adding the pd_prune_xid field (which enlarges the header).
*/
#define PG_PAGE_LAYOUT_VERSION 4
#define PageClearHasFreeLinePointers(page) \
(((PageHeader) (page))->pd_flags &= ~PD_HAS_FREE_LINES)
-#define PageIsPrunable(page) \
- (((PageHeader) (page))->pd_flags & PD_PRUNABLE)
-#define PageSetPrunable(page) \
- (((PageHeader) (page))->pd_flags |= PD_PRUNABLE)
-#define PageClearPrunable(page) \
- (((PageHeader) (page))->pd_flags &= ~PD_PRUNABLE)
-
#define PageIsFull(page) \
(((PageHeader) (page))->pd_flags & PD_PAGE_FULL)
#define PageSetFull(page) \
#define PageClearFull(page) \
(((PageHeader) (page))->pd_flags &= ~PD_PAGE_FULL)
+#define PageIsPrunable(page, oldestxmin) \
+( \
+ AssertMacro(TransactionIdIsNormal(oldestxmin)), \
+ TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \
+ TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \
+)
+#define PageSetPrunable(page, xid) \
+do { \
+ Assert(TransactionIdIsNormal(xid)); \
+ if (!TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) || \
+ TransactionIdPrecedes(xid, ((PageHeader) (page))->pd_prune_xid)) \
+ ((PageHeader) (page))->pd_prune_xid = (xid); \
+} while (0)
+#define PageClearPrunable(page) \
+ (((PageHeader) (page))->pd_prune_xid = InvalidTransactionId)
+
/* ----------------------------------------------------------------
* extern declarations
projects / postgresql / commitdiff