Heikki Linnakangas wrote:
Tom Lane wrote:
I'm not sure whether I like your flag approach better than the
last-used-offset one. The previous patch probably buys some teeny
amount more performance, but the flag seems more robust (noting in
passing that neither patch attempts to WAL-log its changes, so we really
need to treat the values as hints). And a change as sketched here would
leave us 15 free bits for future expansion, which might be nice.
I'll post a patch along those lines.
Here it is.
I'm not fond of the macro names for the flag, but couldn't think of
anything shorter yet descriptive.
--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com
Index: src/backend/access/heap/heapam.c
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/backend/access/heap/heapam.c,v
retrieving revision 1.228
diff -c -r1.228 heapam.c
*** src/backend/access/heap/heapam.c 9 Feb 2007 03:35:33 -0000 1.228
--- src/backend/access/heap/heapam.c 1 Mar 2007 21:29:14 -0000
***************
*** 3327,3332 ****
--- 3327,3335 ----
PageRepairFragmentation(page, NULL);
+ /* clear the hint flag since we just freed some line pointers */
+ HeapPageClearNoFreeLinePointers(page);
+
PageSetLSN(page, lsn);
PageSetTLI(page, ThisTimeLineID);
MarkBufferDirty(buffer);
Index: src/backend/access/heap/hio.c
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/backend/access/heap/hio.c,v
retrieving revision 1.65
diff -c -r1.65 hio.c
*** src/backend/access/heap/hio.c 5 Feb 2007 04:22:18 -0000 1.65
--- src/backend/access/heap/hio.c 1 Mar 2007 21:33:14 -0000
***************
*** 33,51 ****
HeapTuple tuple)
{
Page pageHeader;
! OffsetNumber offnum;
ItemId itemId;
Item item;
- /* Add the tuple to the page */
pageHeader = BufferGetPage(buffer);
offnum = PageAddItem(pageHeader, (Item) tuple->t_data,
! tuple->t_len, InvalidOffsetNumber, LP_USED);
if (offnum == InvalidOffsetNumber)
elog(PANIC, "failed to add tuple to page");
/* Update tuple->t_self to the actual position where it was stored */
ItemPointerSet(&(tuple->t_self), BufferGetBlockNumber(buffer), offnum);
--- 33,65 ----
HeapTuple tuple)
{
Page pageHeader;
! OffsetNumber offnum, maxoff;
ItemId itemId;
Item item;
pageHeader = BufferGetPage(buffer);
+ maxoff = PageGetMaxOffsetNumber(pageHeader);
+
+ /* If we know there's no free line pointers, don't waste cycles
+ * searching for one. We'll set the flag to save work for future
+ * inserts when that happens.
+ */
+ if(HeapPageNoFreeLinePointers(pageHeader))
+ offnum = OffsetNumberNext(maxoff);
+ else
+ offnum = InvalidOffsetNumber;
+
+ /* Add the tuple to the page */
offnum = PageAddItem(pageHeader, (Item) tuple->t_data,
! tuple->t_len, offnum, LP_USED);
if (offnum == InvalidOffsetNumber)
elog(PANIC, "failed to add tuple to page");
+ if(offnum > maxoff)
+ HeapPageSetNoFreeLinePointers(pageHeader);
+
/* Update tuple->t_self to the actual position where it was stored */
ItemPointerSet(&(tuple->t_self), BufferGetBlockNumber(buffer), offnum);
Index: src/backend/commands/vacuum.c
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/backend/commands/vacuum.c,v
retrieving revision 1.346
diff -c -r1.346 vacuum.c
*** src/backend/commands/vacuum.c 15 Feb 2007 23:23:22 -0000 1.346
--- src/backend/commands/vacuum.c 1 Mar 2007 21:41:20 -0000
***************
*** 2453,2458 ****
--- 2453,2460 ----
START_CRIT_SECTION();
uncnt = PageRepairFragmentation(page, unused);
+ if(uncnt > 0)
+ HeapPageClearNoFreeLinePointers(page);
MarkBufferDirty(buf);
***************
*** 2920,2925 ****
--- 2922,2929 ----
}
uncnt = PageRepairFragmentation(page, unused);
+ if(uncnt > 0)
+ HeapPageClearNoFreeLinePointers(page);
MarkBufferDirty(buffer);
Index: src/backend/commands/vacuumlazy.c
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/backend/commands/vacuumlazy.c,v
retrieving revision 1.85
diff -c -r1.85 vacuumlazy.c
*** src/backend/commands/vacuumlazy.c 21 Feb 2007 22:47:45 -0000 1.85
--- src/backend/commands/vacuumlazy.c 1 Mar 2007 21:27:15 -0000
***************
*** 605,610 ****
--- 605,612 ----
}
uncnt = PageRepairFragmentation(page, unused);
+ if(uncnt > 0)
+ HeapPageClearNoFreeLinePointers(page);
MarkBufferDirty(buffer);
Index: src/include/access/htup.h
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/include/access/htup.h,v
retrieving revision 1.92
diff -c -r1.92 htup.h
*** src/include/access/htup.h 27 Feb 2007 23:48:09 -0000 1.92
--- src/include/access/htup.h 1 Mar 2007 22:13:25 -0000
***************
*** 17,22 ****
--- 17,42 ----
#include "storage/itemptr.h"
#include "storage/relfilenode.h"
+ /* We use the pd_amprivate field in the page header to store flags.
+ * At the moment, there's just one flag.
+ */
+
+ /* HP_NOFREELINEPOINTERS flag should be set as a hint when there definitely
+ * isn't any free line pointers left on the page. However, absence of the
+ * flag doesn't necessarily mean that there is a free line pointer.
+ */
+ #define HP_NOFREELINEPOINTERS (1 << 0)
+
+ #define HeapPageNoFreeLinePointers(page) \
+ (((PageHeader)page)->pd_amprivate & HP_NOFREELINEPOINTERS)
+
+ #define HeapPageSetNoFreeLinePointers(page) \
+ (((PageHeader)page)->pd_amprivate |= HP_NOFREELINEPOINTERS)
+
+ #define HeapPageClearNoFreeLinePointers(page) \
+ (((PageHeader)page)->pd_amprivate &= ~HP_NOFREELINEPOINTERS)
+
+
/*
* MaxTupleAttributeNumber limits the number of (user) columns in a tuple.
* The key limit on this value is that the size of the fixed overhead for
Index: src/include/storage/bufpage.h
===================================================================
RCS file: /home/hlinnaka/pgcvsrepository/pgsql/src/include/storage/bufpage.h,v
retrieving revision 1.71
diff -c -r1.71 bufpage.h
*** src/include/storage/bufpage.h 21 Feb 2007 20:02:17 -0000 1.71
--- src/include/storage/bufpage.h 1 Mar 2007 22:14:37 -0000
***************
*** 68,74 ****
* space"; each AM has an "opaque" structure defined somewhere that is
* stored as the page trailer. an access method should always
* initialize its pages with PageInit and then set its own opaque
! * fields.
*/
typedef Pointer Page;
--- 68,76 ----
* space"; each AM has an "opaque" structure defined somewhere that is
* stored as the page trailer. an access method should always
* initialize its pages with PageInit and then set its own opaque
! * fields. In addition to the special space, there's a 16-bit
! * pd_amprivate field in the page header itself reserved for AM-specific
! * use.
*/
typedef Pointer Page;
***************
*** 90,95 ****
--- 92,98 ----
*
* pd_lsn - identifies xlog record for last change to this page.
* pd_tli - ditto.
+ * pd_amprivate - reserved for AM-specific use
* pd_lower - offset to start of free space.
* pd_upper - offset to end of free space.
* pd_special - offset to start of special space.
***************
*** 98,105 ****
* 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
* to disk until xlog has been flushed at least as far as the page's LSN.
! * We also store the TLI for identification purposes (it is not clear that
! * this is actually necessary, but it seems like a good idea).
*
* The page version number and page size are packed together into a single
* uint16 field. This is for historical reasons: before PostgreSQL 7.3,
--- 101,109 ----
* 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
* to disk until xlog has been flushed at least as far as the page's LSN.
! * We also store the 16 least significant bits of the TLI for identification
! * purposes (it is not clear that this is actually necessary, but it seems
! * like a good idea).
*
* The page version number and page size are packed together into a single
* uint16 field. This is for historical reasons: before PostgreSQL 7.3,
***************
*** 119,125 ****
/* XXX LSN is member of *any* block, not only page-organized ones */
XLogRecPtr pd_lsn; /* LSN: next byte after last byte of xlog
* record for last change to this page */
! TimeLineID pd_tli; /* TLI of last change */
LocationIndex pd_lower; /* offset to start of free space */
LocationIndex pd_upper; /* offset to end of free space */
LocationIndex pd_special; /* offset to start of special space */
--- 123,131 ----
/* XXX LSN is member of *any* block, not only page-organized ones */
XLogRecPtr pd_lsn; /* LSN: next byte after last byte of xlog
* record for last change to this page */
! uint16 pd_tli; /* least significant bits of the TLI
! * corresponding the LSN. */
! uint16 pd_amprivate; /* am-specific information */
LocationIndex pd_lower; /* offset to start of free space */
LocationIndex pd_upper; /* offset to end of free space */
LocationIndex pd_special; /* offset to start of special space */
***************
*** 304,313 ****
#define PageSetLSN(page, lsn) \
(((PageHeader) (page))->pd_lsn = (lsn))
#define PageGetTLI(page) \
(((PageHeader) (page))->pd_tli)
#define PageSetTLI(page, tli) \
! (((PageHeader) (page))->pd_tli = (tli))
/* ----------------------------------------------------------------
* extern declarations
--- 310,320 ----
#define PageSetLSN(page, lsn) \
(((PageHeader) (page))->pd_lsn = (lsn))
+ /* NOTE: only the 16 least significant bits are stored */
#define PageGetTLI(page) \
(((PageHeader) (page))->pd_tli)
#define PageSetTLI(page, tli) \
! (((PageHeader) (page))->pd_tli = (tli) & 0x0000FFFF)
/* ----------------------------------------------------------------
* extern declarations
---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings
