More infrastructure for btree compaction project. Tree-traversal code

now knows what to do upon hitting a dead page (in theory anyway, it's
untested...).  Add a post-VACUUM-cleanup entry point for index AMs, to
provide a place for dead-page scavenging to happen.
Also, fix oversight that broke btpo_prev links in temporary indexes.
initdb forced due to additions in pg_am.

3 files changed, 40 insertions, 19 deletions

diff --git a/src/include/access/genam.h b/src/include/access/genam.h
index 6266da47c8f..59ecf1d8f4f 100644
--- a/src/include/access/genam.h
+++ b/src/include/access/genam.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: genam.h,v 1.37 2002/09/04 20:31:36 momjian Exp $
+ * $Id: genam.h,v 1.38 2003/02/22 00:45:05 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -20,17 +20,32 @@
 #include "nodes/primnodes.h"
 
 
-/* Struct for statistics returned by bulk-delete operation */
+/*
+ * Struct for statistics returned by bulk-delete operation
+ *
+ * This is now also passed to the index AM's vacuum-cleanup operation,
+ * if it has one, which can modify the results as needed.  Note that
+ * an index AM could choose to have bulk-delete return a larger struct
+ * of which this is just the first field; this provides a way for bulk-delete
+ * to communicate additional private data to vacuum-cleanup.
+ */
 typedef struct IndexBulkDeleteResult
 {
 	BlockNumber num_pages;		/* pages remaining in index */
+	double		num_index_tuples;		/* tuples remaining */
 	double		tuples_removed; /* # removed by bulk-delete operation */
-	double		num_index_tuples;		/* # remaining */
+	BlockNumber	pages_free;		/* # unused pages in index */
 } IndexBulkDeleteResult;
 
 /* Typedef for callback function to determine if a tuple is bulk-deletable */
 typedef bool (*IndexBulkDeleteCallback) (ItemPointer itemptr, void *state);
 
+/* Struct for additional arguments passed to vacuum-cleanup operation */
+typedef struct IndexVacuumCleanupInfo
+{
+	bool		vacuum_full;	/* VACUUM FULL (we have exclusive lock) */
+	int			message_level;	/* elog level for progress messages */
+} IndexVacuumCleanupInfo;
 
 /* Struct for heap-or-index scans of system tables */
 typedef struct SysScanDescData
@@ -72,6 +87,9 @@ extern bool index_getnext_indexitem(IndexScanDesc scan,
 extern IndexBulkDeleteResult *index_bulk_delete(Relation indexRelation,
 				  IndexBulkDeleteCallback callback,
 				  void *callback_state);
+extern IndexBulkDeleteResult *index_vacuum_cleanup(Relation indexRelation,
+				  IndexVacuumCleanupInfo *info,
+				  IndexBulkDeleteResult *stats);
 extern RegProcedure index_cost_estimator(Relation indexRelation);
 extern RegProcedure index_getprocid(Relation irel, AttrNumber attnum,
 				uint16 procnum);
diff --git a/src/include/access/nbtree.h b/src/include/access/nbtree.h
index f4dce1842f1..4bb5db0513e 100644
--- a/src/include/access/nbtree.h
+++ b/src/include/access/nbtree.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: nbtree.h,v 1.64 2003/02/21 00:06:22 tgl Exp $
+ * $Id: nbtree.h,v 1.65 2003/02/22 00:45:05 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -54,6 +54,7 @@ typedef BTPageOpaqueData *BTPageOpaque;
 #define BTP_ROOT		(1 << 1)	/* root page (has no parent) */
 #define BTP_DELETED		(1 << 2)	/* page has been deleted from tree */
 #define BTP_META		(1 << 3)	/* meta-page */
+#define BTP_HALF_DEAD	(1 << 4)	/* empty, but still in tree */
 
 
 /*
@@ -124,12 +125,13 @@ typedef BTItemData *BTItem;
 #define SizeOfBTItem	sizeof(BTItemData)
 
 /* Test whether items are the "same" per the above notes */
-#define BTItemSame(i1, i2)	  ( (i1)->bti_itup.t_tid.ip_blkid.bi_hi == \
-								(i2)->bti_itup.t_tid.ip_blkid.bi_hi && \
-								(i1)->bti_itup.t_tid.ip_blkid.bi_lo == \
-								(i2)->bti_itup.t_tid.ip_blkid.bi_lo && \
-								(i1)->bti_itup.t_tid.ip_posid == \
-								(i2)->bti_itup.t_tid.ip_posid )
+#define BTTidSame(i1, i2)	\
+	( (i1).ip_blkid.bi_hi == (i2).ip_blkid.bi_hi && \
+	  (i1).ip_blkid.bi_lo == (i2).ip_blkid.bi_lo && \
+	  (i1).ip_posid == (i2).ip_posid )
+#define BTItemSame(i1, i2)	\
+	BTTidSame((i1)->bti_itup.t_tid, (i2)->bti_itup.t_tid)
+
 
 /*
  *	In general, the btree code tries to localize its knowledge about
@@ -150,6 +152,7 @@ typedef BTItemData *BTItem;
 #define P_ISLEAF(opaque)		((opaque)->btpo_flags & BTP_LEAF)
 #define P_ISROOT(opaque)		((opaque)->btpo_flags & BTP_ROOT)
 #define P_ISDELETED(opaque)		((opaque)->btpo_flags & BTP_DELETED)
+#define P_IGNORE(opaque)		((opaque)->btpo_flags & (BTP_DELETED|BTP_HALF_DEAD))
 
 /*
  *	Lehman and Yao's algorithm requires a ``high key'' on every non-rightmost
@@ -412,8 +415,6 @@ typedef BTScanOpaqueData *BTScanOpaque;
 /*
  * prototypes for functions in nbtree.c (external entry points for btree)
  */
-extern bool BuildingBtree;		/* in nbtree.c */
-
 extern void AtEOXact_nbtree(void);
 
 extern Datum btbuild(PG_FUNCTION_ARGS);
@@ -426,6 +427,7 @@ extern Datum btendscan(PG_FUNCTION_ARGS);
 extern Datum btmarkpos(PG_FUNCTION_ARGS);
 extern Datum btrestrpos(PG_FUNCTION_ARGS);
 extern Datum btbulkdelete(PG_FUNCTION_ARGS);
+extern Datum btvacuumcleanup(PG_FUNCTION_ARGS);
 
 /*
  * prototypes for functions in nbtinsert.c
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index a1be9bacf3b..cb2e6e523df 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2002, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: xlog.h,v 1.41 2003/02/21 00:06:22 tgl Exp $
+ * $Id: xlog.h,v 1.42 2003/02/22 00:45:05 tgl Exp $
  */
 #ifndef XLOG_H
 #define XLOG_H
@@ -56,17 +56,18 @@ typedef struct XLogRecord
 #define XLR_INFO_MASK			0x0F
 
 /*
- * We support backup of up to 2 disk blocks per XLOG record (could support
- * more if we cared to dedicate more xl_info bits for this purpose; currently
- * do not need more than 2 anyway).  If we backed up any disk blocks then we
- * use flag bits in xl_info to signal it.
+ * If we backed up any disk blocks with the XLOG record, we use flag bits in
+ * xl_info to signal it.  We support backup of up to 3 disk blocks per XLOG
+ * record.  (Could support 4 if we cared to dedicate all the xl_info bits for
+ * this purpose; currently bit 0 of xl_info is unused and available.)
  */
-#define XLR_BKP_BLOCK_MASK		0x0C	/* all info bits used for bkp
+#define XLR_BKP_BLOCK_MASK		0x0E	/* all info bits used for bkp
 										 * blocks */
-#define XLR_MAX_BKP_BLOCKS		2
+#define XLR_MAX_BKP_BLOCKS		3
 #define XLR_SET_BKP_BLOCK(iblk) (0x08 >> (iblk))
 #define XLR_BKP_BLOCK_1			XLR_SET_BKP_BLOCK(0)	/* 0x08 */
 #define XLR_BKP_BLOCK_2			XLR_SET_BKP_BLOCK(1)	/* 0x04 */
+#define XLR_BKP_BLOCK_3			XLR_SET_BKP_BLOCK(2)	/* 0x02 */
 
 /*
  * Sometimes we log records which are out of transaction control.