Make large sequential scans and VACUUMs work in a limited-size "ring" of

buffers, rather than blowing out the whole shared-buffer arena.  Aside from
avoiding cache spoliation, this fixes the problem that VACUUM formerly tended
to cause a WAL flush for every page it modified, because we had it hacked to
use only a single buffer.  Those flushes will now occur only once per
ring-ful.  The exact ring size, and the threshold for seqscans to switch into
the ring usage pattern, remain under debate; but the infrastructure seems
done.  The key bit of infrastructure is a new optional BufferAccessStrategy
object that can be passed to ReadBuffer operations; this replaces the former
StrategyHintVacuum API.

This patch also changes the buffer usage-count methodology a bit: we now
advance usage_count when first pinning a buffer, rather than when last
unpinning it.  To preserve the behavior that a buffer's lifetime starts to
decrease when it's released, the clock sweep code is modified to not decrement
usage_count of pinned buffers.

Work not done in this commit: teach GiST and GIN indexes to use the vacuum
BufferAccessStrategy for vacuum-driven fetches.

Original patch by Simon, reworked by Heikki and again by Tom.

8 files changed, 51 insertions, 19 deletions

diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index 1f31baf0e45..98296e62bef 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -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/genam.h,v 1.66 2007/01/05 22:19:50 momjian Exp $
+ * $PostgreSQL: pgsql/src/include/access/genam.h,v 1.67 2007/05/30 20:12:02 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -40,6 +40,7 @@ typedef struct IndexVacuumInfo
 	bool		vacuum_full;	/* VACUUM FULL (we have exclusive lock) */
 	int			message_level;	/* ereport level for progress messages */
 	double		num_heap_tuples;	/* tuples remaining in heap */
+	BufferAccessStrategy strategy;	/* access strategy for reads */
 } IndexVacuumInfo;
 
 /*
diff --git a/src/include/access/hash.h b/src/include/access/hash.h
index d382ee6ee91..2bd314a8aa3 100644
--- a/src/include/access/hash.h
+++ b/src/include/access/hash.h
@@ -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 1.80 2007/05/03 16:45:58 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/hash.h,v 1.81 2007/05/30 20:12:02 tgl Exp $
  *
  * NOTES
  *		modeled after Margo Seltzer's hash implementation for unix.
@@ -273,11 +273,13 @@ extern void _hash_doinsert(Relation rel, IndexTuple itup);
 
 /* hashovfl.c */
 extern Buffer _hash_addovflpage(Relation rel, Buffer metabuf, Buffer buf);
-extern BlockNumber _hash_freeovflpage(Relation rel, Buffer ovflbuf);
+extern BlockNumber _hash_freeovflpage(Relation rel, Buffer ovflbuf,
+									  BufferAccessStrategy bstrategy);
 extern void _hash_initbitmap(Relation rel, HashMetaPage metap,
 				 BlockNumber blkno);
 extern void _hash_squeezebucket(Relation rel,
-					Bucket bucket, BlockNumber bucket_blkno);
+								Bucket bucket, BlockNumber bucket_blkno,
+								BufferAccessStrategy bstrategy);
 
 /* hashpage.c */
 extern void _hash_getlock(Relation rel, BlockNumber whichlock, int access);
@@ -287,6 +289,9 @@ extern Buffer _hash_getbuf(Relation rel, BlockNumber blkno,
 						   int access, int flags);
 extern Buffer _hash_getinitbuf(Relation rel, BlockNumber blkno);
 extern Buffer _hash_getnewbuf(Relation rel, BlockNumber blkno);
+extern Buffer _hash_getbuf_with_strategy(Relation rel, BlockNumber blkno,
+										 int access, int flags,
+										 BufferAccessStrategy bstrategy);
 extern void _hash_relbuf(Relation rel, Buffer buf);
 extern void _hash_dropbuf(Relation rel, Buffer buf);
 extern void _hash_wrtbuf(Relation rel, Buffer buf);
diff --git a/src/include/access/relscan.h b/src/include/access/relscan.h
index 7a1ea39352a..200b45713e7 100644
--- a/src/include/access/relscan.h
+++ b/src/include/access/relscan.h
@@ -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/relscan.h,v 1.53 2007/05/27 03:50:39 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/relscan.h,v 1.54 2007/05/30 20:12:02 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -27,6 +27,7 @@ typedef struct HeapScanDescData
 	int			rs_nkeys;		/* number of scan keys */
 	ScanKey		rs_key;			/* array of scan key descriptors */
 	BlockNumber rs_nblocks;		/* number of blocks to scan */
+	BufferAccessStrategy rs_strategy;	/* access strategy for reads */
 	bool		rs_pageatatime; /* verify visibility page-at-a-time? */
 
 	/* scan current state */
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 87ff6aba501..1c741f38fd0 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/xlog.h,v 1.77 2007/05/20 21:08:19 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/xlog.h,v 1.78 2007/05/30 20:12:02 tgl Exp $
  */
 #ifndef XLOG_H
 #define XLOG_H
@@ -159,6 +159,7 @@ extern bool XLOG_DEBUG;
 
 extern XLogRecPtr XLogInsert(RmgrId rmid, uint8 info, XLogRecData *rdata);
 extern void XLogFlush(XLogRecPtr RecPtr);
+extern bool XLogNeedsFlush(XLogRecPtr RecPtr);
 
 extern void xlog_redo(XLogRecPtr lsn, XLogRecord *record);
 extern void xlog_desc(StringInfo buf, uint8 xl_info, char *rec);
diff --git a/src/include/commands/vacuum.h b/src/include/commands/vacuum.h
index acb2f623e27..50a475bc5e3 100644
--- a/src/include/commands/vacuum.h
+++ b/src/include/commands/vacuum.h
@@ -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/commands/vacuum.h,v 1.71 2007/05/17 15:28:29 alvherre Exp $
+ * $PostgreSQL: pgsql/src/include/commands/vacuum.h,v 1.72 2007/05/30 20:12:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -18,9 +18,11 @@
 #include "catalog/pg_statistic.h"
 #include "catalog/pg_type.h"
 #include "nodes/parsenodes.h"
+#include "storage/buf.h"
 #include "storage/lock.h"
 #include "utils/rel.h"
 
+
 /*----------
  * ANALYZE builds one of these structs for each attribute (column) that is
  * to be analyzed.	The struct and subsidiary data are in anl_context,
@@ -110,7 +112,8 @@ extern int	vacuum_freeze_min_age;
 
 
 /* in commands/vacuum.c */
-extern void vacuum(VacuumStmt *vacstmt, List *relids, bool isTopLevel);
+extern void vacuum(VacuumStmt *vacstmt, List *relids,
+				   BufferAccessStrategy bstrategy, bool isTopLevel);
 extern void vac_open_indexes(Relation relation, LOCKMODE lockmode,
 				 int *nindexes, Relation **Irel);
 extern void vac_close_indexes(int nindexes, Relation *Irel, LOCKMODE lockmode);
@@ -127,9 +130,11 @@ extern bool vac_is_partial_index(Relation indrel);
 extern void vacuum_delay_point(void);
 
 /* in commands/vacuumlazy.c */
-extern void lazy_vacuum_rel(Relation onerel, VacuumStmt *vacstmt);
+extern void lazy_vacuum_rel(Relation onerel, VacuumStmt *vacstmt,
+							BufferAccessStrategy bstrategy);
 
 /* in commands/analyze.c */
-extern void analyze_rel(Oid relid, VacuumStmt *vacstmt);
+extern void analyze_rel(Oid relid, VacuumStmt *vacstmt,
+						BufferAccessStrategy bstrategy);
 
 #endif   /* VACUUM_H */
diff --git a/src/include/storage/buf.h b/src/include/storage/buf.h
index 94da564d1eb..a812a9e269a 100644
--- a/src/include/storage/buf.h
+++ b/src/include/storage/buf.h
@@ -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/storage/buf.h,v 1.21 2007/01/05 22:19:57 momjian Exp $
+ * $PostgreSQL: pgsql/src/include/storage/buf.h,v 1.22 2007/05/30 20:12:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -36,4 +36,11 @@ typedef int Buffer;
  */
 #define BufferIsLocal(buffer)	((buffer) < 0)
 
+/*
+ * Buffer access strategy objects.
+ *
+ * BufferAccessStrategyData is private to freelist.c
+ */
+typedef struct BufferAccessStrategyData *BufferAccessStrategy;
+
 #endif   /* BUF_H */
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 561b7e40f69..d5eef8734ff 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -8,7 +8,7 @@
  * Portions Copyright (c) 1996-2007, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/storage/buf_internals.h,v 1.89 2007/01/05 22:19:57 momjian Exp $
+ * $PostgreSQL: pgsql/src/include/storage/buf_internals.h,v 1.90 2007/05/30 20:12:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -167,9 +167,6 @@ extern DLLIMPORT BufferDesc *BufferDescriptors;
 /* in localbuf.c */
 extern BufferDesc *LocalBufferDescriptors;
 
-/* in freelist.c */
-extern bool strategy_hint_vacuum;
-
 /* event counters in buf_init.c */
 extern long int ReadBufferCount;
 extern long int ReadLocalBufferCount;
@@ -184,8 +181,12 @@ extern long int LocalBufferFlushCount;
  */
 
 /* freelist.c */
-extern volatile BufferDesc *StrategyGetBuffer(void);
-extern void StrategyFreeBuffer(volatile BufferDesc *buf, bool at_head);
+extern volatile BufferDesc *StrategyGetBuffer(BufferAccessStrategy strategy,
+											  bool *lock_held);
+extern void StrategyFreeBuffer(volatile BufferDesc *buf);
+extern bool StrategyRejectBuffer(BufferAccessStrategy strategy,
+								 volatile BufferDesc *buf);
+
 extern int	StrategySyncStart(void);
 extern Size StrategyShmemSize(void);
 extern void StrategyInitialize(bool init);
diff --git a/src/include/storage/bufmgr.h b/src/include/storage/bufmgr.h
index ad203621793..9ae83b4253e 100644
--- a/src/include/storage/bufmgr.h
+++ b/src/include/storage/bufmgr.h
@@ -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/storage/bufmgr.h,v 1.103 2007/05/02 23:18:03 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/storage/bufmgr.h,v 1.104 2007/05/30 20:12:03 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -19,6 +19,14 @@
 
 typedef void *Block;
 
+/* Possible arguments for GetAccessStrategy() */
+typedef enum BufferAccessStrategyType
+{
+	BAS_NORMAL,		/* Normal random access */
+	BAS_BULKREAD,	/* Large read-only scan (hint bit updates are ok) */
+	BAS_VACUUM		/* VACUUM */
+} BufferAccessStrategyType;
+
 /* in globals.c ... this duplicates miscadmin.h */
 extern DLLIMPORT int NBuffers;
 
@@ -111,6 +119,8 @@ extern DLLIMPORT int32 *LocalRefCount;
  * prototypes for functions in bufmgr.c
  */
 extern Buffer ReadBuffer(Relation reln, BlockNumber blockNum);
+extern Buffer ReadBufferWithStrategy(Relation reln, BlockNumber blockNum,
+									 BufferAccessStrategy strategy);
 extern Buffer ReadOrZeroBuffer(Relation reln, BlockNumber blockNum);
 extern void ReleaseBuffer(Buffer buffer);
 extern void UnlockReleaseBuffer(Buffer buffer);
@@ -157,6 +167,7 @@ extern void BgBufferSync(void);
 extern void AtProcExit_LocalBuffers(void);
 
 /* in freelist.c */
-extern void StrategyHintVacuum(bool vacuum_active);
+extern BufferAccessStrategy GetAccessStrategy(BufferAccessStrategyType btype);
+extern void FreeAccessStrategy(BufferAccessStrategy strategy);
 
 #endif