Nested transactions. There is still much left to do, especially on the

performance front, but with feature freeze upon us I think it's time to
drive a stake in the ground and say that this will be in 7.5.

Alvaro Herrera, with some help from Tom Lane.

11 files changed, 119 insertions, 74 deletions

diff --git a/src/include/access/clog.h b/src/include/access/clog.h
index bd7f4152be3..2df1cedc1c9 100644
--- a/src/include/access/clog.h
+++ b/src/include/access/clog.h
@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/clog.h,v 1.8 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/clog.h,v 1.9 2004/07/01 00:51:38 tgl Exp $
  */
 #ifndef CLOG_H
 #define CLOG_H
@@ -16,13 +16,16 @@
 /*
  * Possible transaction statuses --- note that all-zeroes is the initial
  * state.
+ *
+ * A "subcommitted" transaction is a committed subtransaction whose parent
+ * hasn't committed or aborted yet.
  */
 typedef int XidStatus;
 
 #define TRANSACTION_STATUS_IN_PROGRESS		0x00
 #define TRANSACTION_STATUS_COMMITTED		0x01
 #define TRANSACTION_STATUS_ABORTED			0x02
-/* 0x03 is available without changing commit log space allocation */
+#define TRANSACTION_STATUS_SUB_COMMITTED	0x03
 
 /* exported because lwlock.c needs it */
 #define NUM_CLOG_BUFFERS	8
@@ -39,12 +42,6 @@ extern void ShutdownCLOG(void);
 extern void CheckPointCLOG(void);
 extern void ExtendCLOG(TransactionId newestXact);
 extern void TruncateCLOG(TransactionId oldestXact);
-
-/* XLOG stuff */
-#define CLOG_ZEROPAGE		0x00
-
-extern void clog_redo(XLogRecPtr lsn, XLogRecord *record);
-extern void clog_undo(XLogRecPtr lsn, XLogRecord *record);
-extern void clog_desc(char *buf, uint8 xl_info, char *rec);
+extern void clog_zeropage_redo(int pageno);
 
 #endif   /* CLOG_H */
diff --git a/src/include/access/gistscan.h b/src/include/access/gistscan.h
index b8466429959..4022f542752 100644
--- a/src/include/access/gistscan.h
+++ b/src/include/access/gistscan.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/gistscan.h,v 1.22 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/gistscan.h,v 1.23 2004/07/01 00:51:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -23,5 +23,6 @@ extern Datum gistrestrpos(PG_FUNCTION_ARGS);
 extern Datum gistendscan(PG_FUNCTION_ARGS);
 extern void gistadjscans(Relation r, int op, BlockNumber blkno, OffsetNumber offnum);
 extern void AtEOXact_gist(void);
+extern void AtEOSubXact_gist(TransactionId childXid);
 
 #endif   /* GISTSCAN_H */
diff --git a/src/include/access/hash.h b/src/include/access/hash.h
index ffeea63417b..2088cc2f5a6 100644
--- a/src/include/access/hash.h
+++ b/src/include/access/hash.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/hash.h,v 1.54 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/hash.h,v 1.55 2004/07/01 00:51:38 tgl Exp $
  *
  * NOTES
  *		modeled after Margo Seltzer's hash implementation for unix.
@@ -293,6 +293,7 @@ extern void _hash_regscan(IndexScanDesc scan);
 extern void _hash_dropscan(IndexScanDesc scan);
 extern bool _hash_has_active_scan(Relation rel, Bucket bucket);
 extern void AtEOXact_hash(void);
+extern void AtEOSubXact_hash(TransactionId childXid);
 
 /* hashsearch.c */
 extern bool _hash_next(IndexScanDesc scan, ScanDirection dir);
diff --git a/src/include/access/htup.h b/src/include/access/htup.h
index 3d48b5f45a3..fdcfc8dc6f1 100644
--- a/src/include/access/htup.h
+++ b/src/include/access/htup.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/htup.h,v 1.65 2004/04/01 21:28:45 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/htup.h,v 1.66 2004/07/01 00:51:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -109,18 +109,14 @@
 typedef struct HeapTupleFields
 {
 	TransactionId t_xmin;		/* inserting xact ID */
-
-	union
-	{
-		CommandId	t_cmin;		/* inserting command ID */
-		TransactionId t_xmax;	/* deleting xact ID */
-	}			t_field2;
+	CommandId	t_cmin;			/* inserting command ID */
+	TransactionId t_xmax;		/* deleting xact ID */
 
 	union
 	{
 		CommandId	t_cmax;		/* deleting command ID */
 		TransactionId t_xvac;	/* VACUUM FULL xact ID */
-	}			t_field3;
+	}			t_field4;
 } HeapTupleFields;
 
 typedef struct DatumTupleFields
@@ -172,9 +168,7 @@ typedef HeapTupleHeaderData *HeapTupleHeader;
 										 * attribute(s) */
 #define HEAP_HASEXTENDED		0x000C	/* the two above combined */
 #define HEAP_HASOID				0x0010	/* has an object-id field */
-/* bit 0x0020 is presently unused */
-#define HEAP_XMAX_IS_XMIN		0x0040	/* created and deleted in the same
-										 * transaction */
+/* 0x0020 and 0x0040 are unused */
 #define HEAP_XMAX_UNLOGGED		0x0080	/* to lock tuple for update
 										 * without logging */
 #define HEAP_XMIN_COMMITTED		0x0100	/* t_xmin committed */
@@ -211,62 +205,47 @@ typedef HeapTupleHeaderData *HeapTupleHeader;
 
 #define HeapTupleHeaderGetXmax(tup) \
 ( \
-	((tup)->t_infomask & HEAP_XMAX_IS_XMIN) ? \
-		(tup)->t_choice.t_heap.t_xmin \
-	: \
-		(tup)->t_choice.t_heap.t_field2.t_xmax \
+	(tup)->t_choice.t_heap.t_xmax \
 )
 
 #define HeapTupleHeaderSetXmax(tup, xid) \
-do { \
-	TransactionId	_newxid = (xid); \
-	if (TransactionIdEquals((tup)->t_choice.t_heap.t_xmin, _newxid)) \
-		(tup)->t_infomask |= HEAP_XMAX_IS_XMIN; \
-	else \
-	{ \
-		(tup)->t_infomask &= ~HEAP_XMAX_IS_XMIN; \
-		TransactionIdStore(_newxid, &(tup)->t_choice.t_heap.t_field2.t_xmax); \
-	} \
-} while (0)
+( \
+	TransactionIdStore((xid), &(tup)->t_choice.t_heap.t_xmax) \
+)
 
-/*
- * Note: GetCmin will produce wrong answers after SetXmax has been executed
- * by a transaction other than the inserting one.  We could check
- * HEAP_XMAX_INVALID and return FirstCommandId if it's clear, but since that
- * bit will be set again if the deleting transaction aborts, there'd be no
- * real gain in safety from the extra test.  So, just rely on the caller not
- * to trust the value unless it's meaningful.
- */
 #define HeapTupleHeaderGetCmin(tup) \
 ( \
-	(tup)->t_choice.t_heap.t_field2.t_cmin \
+	(tup)->t_choice.t_heap.t_cmin \
 )
 
 #define HeapTupleHeaderSetCmin(tup, cid) \
-do { \
-	Assert((tup)->t_infomask & HEAP_XMAX_INVALID); \
-	(tup)->t_choice.t_heap.t_field2.t_cmin = (cid); \
-} while (0)
+( \
+	(tup)->t_choice.t_heap.t_cmin = (cid) \
+)
 
 /*
- * As with GetCmin, we can't completely ensure that GetCmax can detect whether
- * a valid command ID is available, and there's little point in a partial test.
+ * Note: GetCmax will produce wrong answers after SetXvac has been executed
+ * by a transaction other than the inserting one.  We could check
+ * HEAP_XMAX_INVALID and return FirstCommandId if it's clear, but since that
+ * bit will be set again if the deleting transaction aborts, there'd be no
+ * real gain in safety from the extra test.  So, just rely on the caller not
+ * to trust the value unless it's meaningful.
  */
 #define HeapTupleHeaderGetCmax(tup) \
 ( \
-	(tup)->t_choice.t_heap.t_field3.t_cmax \
+	(tup)->t_choice.t_heap.t_field4.t_cmax \
 )
 
 #define HeapTupleHeaderSetCmax(tup, cid) \
 do { \
 	Assert(!((tup)->t_infomask & HEAP_MOVED)); \
-	(tup)->t_choice.t_heap.t_field3.t_cmax = (cid); \
+	(tup)->t_choice.t_heap.t_field4.t_cmax = (cid); \
 } while (0)
 
 #define HeapTupleHeaderGetXvac(tup) \
 ( \
 	((tup)->t_infomask & HEAP_MOVED) ? \
-		(tup)->t_choice.t_heap.t_field3.t_xvac \
+		(tup)->t_choice.t_heap.t_field4.t_xvac \
 	: \
 		InvalidTransactionId \
 )
@@ -274,7 +253,7 @@ do { \
 #define HeapTupleHeaderSetXvac(tup, xid) \
 do { \
 	Assert((tup)->t_infomask & HEAP_MOVED); \
-	TransactionIdStore((xid), &(tup)->t_choice.t_heap.t_field3.t_xvac); \
+	TransactionIdStore((xid), &(tup)->t_choice.t_heap.t_field4.t_xvac); \
 } while (0)
 
 #define HeapTupleHeaderGetDatumLength(tup) \
diff --git a/src/include/access/rmgr.h b/src/include/access/rmgr.h
index e63e3fbc31d..7ea3134031d 100644
--- a/src/include/access/rmgr.h
+++ b/src/include/access/rmgr.h
@@ -3,7 +3,7 @@
  *
  * Resource managers definition
  *
- * $PostgreSQL: pgsql/src/include/access/rmgr.h,v 1.10 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/rmgr.h,v 1.11 2004/07/01 00:51:38 tgl Exp $
  */
 #ifndef RMGR_H
 #define RMGR_H
@@ -16,7 +16,7 @@ typedef uint8 RmgrId;
 #define RM_XLOG_ID				0
 #define RM_XACT_ID				1
 #define RM_SMGR_ID				2
-#define RM_CLOG_ID				3
+#define RM_SLRU_ID				3
 #define RM_HEAP_ID				10
 #define RM_BTREE_ID				11
 #define RM_HASH_ID				12
diff --git a/src/include/access/rtree.h b/src/include/access/rtree.h
index fdb33eba06f..5b5347e9a08 100644
--- a/src/include/access/rtree.h
+++ b/src/include/access/rtree.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/rtree.h,v 1.32 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/rtree.h,v 1.33 2004/07/01 00:51:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -130,6 +130,7 @@ extern void rtree_desc(char *buf, uint8 xl_info, char *rec);
 extern void rtadjscans(Relation r, int op, BlockNumber blkno,
 		   OffsetNumber offnum);
 extern void AtEOXact_rtree(void);
+extern void AtEOSubXact_rtree(TransactionId childXid);
 
 /* rtstrat.c */
 extern StrategyNumber RTMapToInternalOperator(StrategyNumber strat);
diff --git a/src/include/access/slru.h b/src/include/access/slru.h
index 213cca5c216..e3245fac658 100644
--- a/src/include/access/slru.h
+++ b/src/include/access/slru.h
@@ -6,11 +6,12 @@
  * Portions Copyright (c) 2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/slru.h,v 1.6 2004/05/31 03:48:08 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/slru.h,v 1.7 2004/07/01 00:51:38 tgl Exp $
  */
 #ifndef SLRU_H
 #define SLRU_H
 
+#include "access/xlog.h"
 #include "storage/lwlock.h"
 
 
@@ -56,4 +57,12 @@ extern void SimpleLruSetLatestPage(SlruCtl ctl, int pageno);
 extern void SimpleLruFlush(SlruCtl ctl, bool checkpoint);
 extern void SimpleLruTruncate(SlruCtl ctl, int cutoffPage);
 
+/* XLOG stuff */
+#define CLOG_ZEROPAGE		0x00
+#define SUBTRANS_ZEROPAGE	0x10
+
+extern void slru_redo(XLogRecPtr lsn, XLogRecord *record);
+extern void slru_undo(XLogRecPtr lsn, XLogRecord *record);
+extern void slru_desc(char *buf, uint8 xl_info, char *rec);
+
 #endif   /* SLRU_H */
diff --git a/src/include/access/subtrans.h b/src/include/access/subtrans.h
new file mode 100644
index 00000000000..2c601752d12
--- /dev/null
+++ b/src/include/access/subtrans.h
@@ -0,0 +1,35 @@
+/*
+ * subtrans.h
+ *
+ * PostgreSQL subtrans-log manager
+ *
+ * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * $PostgreSQL: pgsql/src/include/access/subtrans.h,v 1.1 2004/07/01 00:51:38 tgl Exp $
+ */
+#ifndef SUBTRANS_H
+#define SUBTRANS_H
+
+#include "access/xlog.h"
+
+/* exported because lwlock.c needs it */
+/* cannot be different from NUM_CLOG_BUFFERS without slru.c changes */
+#define NUM_SUBTRANS_BUFFERS	NUM_CLOG_BUFFERS
+
+extern void SubTransSetParent(TransactionId xid, TransactionId parent);
+extern TransactionId SubTransGetParent(TransactionId xid);
+extern TransactionId SubTransGetTopmostTransaction(TransactionId xid);
+extern bool SubTransXidsHaveCommonAncestor(TransactionId xid1, TransactionId xid2);
+
+extern int	SUBTRANSShmemSize(void);
+extern void SUBTRANSShmemInit(void);
+extern void BootStrapSUBTRANS(void);
+extern void StartupSUBTRANS(void);
+extern void ShutdownSUBTRANS(void);
+extern void CheckPointSUBTRANS(void);
+extern void ExtendSUBTRANS(TransactionId newestXact);
+extern void TruncateSUBTRANS(TransactionId oldestXact);
+extern void subtrans_zeropage_redo(int pageno);
+
+#endif   /* SUBTRANS_H */
diff --git a/src/include/access/transam.h b/src/include/access/transam.h
index 3a2cad7bb03..44e0ff6ea75 100644
--- a/src/include/access/transam.h
+++ b/src/include/access/transam.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/transam.h,v 1.48 2003/11/29 22:40:55 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/access/transam.h,v 1.49 2004/07/01 00:51:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -107,13 +107,16 @@ extern bool TransactionIdDidCommit(TransactionId transactionId);
 extern bool TransactionIdDidAbort(TransactionId transactionId);
 extern void TransactionIdCommit(TransactionId transactionId);
 extern void TransactionIdAbort(TransactionId transactionId);
+extern void TransactionIdSubCommit(TransactionId transactionId);
+extern void TransactionIdCommitTree(int nxids, TransactionId *xids);
+extern void TransactionIdAbortTree(int nxids, TransactionId *xids);
 extern bool TransactionIdPrecedes(TransactionId id1, TransactionId id2);
 extern bool TransactionIdPrecedesOrEquals(TransactionId id1, TransactionId id2);
 extern bool TransactionIdFollows(TransactionId id1, TransactionId id2);
 extern bool TransactionIdFollowsOrEquals(TransactionId id1, TransactionId id2);
 
 /* in transam/varsup.c */
-extern TransactionId GetNewTransactionId(void);
+extern TransactionId GetNewTransactionId(bool isSubXact);
 extern TransactionId ReadNewTransactionId(void);
 extern Oid	GetNewObjectId(void);
 extern void CheckMaxObjectId(Oid assigned_oid);
diff --git a/src/include/access/xact.h b/src/include/access/xact.h
index 53a585ec694..c5b66afd0df 100644
--- a/src/include/access/xact.h
+++ b/src/include/access/xact.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/xact.h,v 1.63 2004/05/22 23:14:38 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/xact.h,v 1.64 2004/07/01 00:51:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -63,7 +63,15 @@ typedef enum TBlockState
 	TBLOCK_INPROGRESS,
 	TBLOCK_END,
 	TBLOCK_ABORT,
-	TBLOCK_ENDABORT
+	TBLOCK_ENDABORT,
+
+	TBLOCK_SUBBEGIN,
+	TBLOCK_SUBBEGINABORT,
+	TBLOCK_SUBINPROGRESS,
+	TBLOCK_SUBEND,
+	TBLOCK_SUBABORT,
+	TBLOCK_SUBENDABORT_OK,
+	TBLOCK_SUBENDABORT_ERROR
 } TBlockState;
 
 /*
@@ -76,12 +84,15 @@ typedef void (*EOXactCallback) (bool isCommit, void *arg);
  */
 typedef struct TransactionStateData
 {
-	TransactionId	transactionIdData;
-	CommandId		commandId;
-	AbsoluteTime	startTime;
-	int				startTimeUsec;
-	TransState		state;
-	TBlockState		blockState;
+	TransactionId	transactionIdData;		/* my XID */
+	CommandId		commandId;				/* current CID */
+	TransState		state;					/* low-level state */
+	TBlockState		blockState;				/* high-level state */
+	int				nestingLevel;			/* nest depth */
+	MemoryContext	curTransactionContext;	/* my xact-lifetime context */
+	List		   *childXids;				/* subcommitted child XIDs */
+	AclId			currentUser;			/* subxact start current_user */
+	struct TransactionStateData *parent;	/* back link to parent */
 } TransactionStateData;
 
 typedef TransactionStateData *TransactionState;
@@ -102,9 +113,11 @@ typedef TransactionStateData *TransactionState;
 typedef struct xl_xact_commit
 {
 	time_t		xtime;
+	int			nrels;			/* number of RelFileNodes */
+	int			nsubxacts;		/* number of subtransaction XIDs */
 	/* Array of RelFileNode(s) to drop at commit */
-	/* The XLOG record length determines how many there are */
 	RelFileNode	xnodes[1];		/* VARIABLE LENGTH ARRAY */
+	/* ARRAY OF COMMITTED SUBTRANSACTION XIDs FOLLOWS */
 } xl_xact_commit;
 
 #define MinSizeOfXactCommit	offsetof(xl_xact_commit, xnodes)
@@ -112,9 +125,11 @@ typedef struct xl_xact_commit
 typedef struct xl_xact_abort
 {
 	time_t		xtime;
+	int			nrels;			/* number of RelFileNodes */
+	int			nsubxacts;		/* number of subtransaction XIDs */
 	/* Array of RelFileNode(s) to drop at abort */
-	/* The XLOG record length determines how many there are */
 	RelFileNode	xnodes[1];		/* VARIABLE LENGTH ARRAY */
+	/* ARRAY OF ABORTED SUBTRANSACTION XIDs FOLLOWS */
 } xl_xact_abort;
 
 #define MinSizeOfXactAbort offsetof(xl_xact_abort, xnodes)
@@ -126,18 +141,20 @@ typedef struct xl_xact_abort
  */
 extern bool IsTransactionState(void);
 extern bool IsAbortedTransactionBlockState(void);
+extern TransactionId GetTopTransactionId(void);
 extern TransactionId GetCurrentTransactionId(void);
 extern CommandId GetCurrentCommandId(void);
 extern AbsoluteTime GetCurrentTransactionStartTime(void);
 extern AbsoluteTime GetCurrentTransactionStartTimeUsec(int *usec);
+extern int	GetCurrentTransactionNestLevel(void);
 extern bool TransactionIdIsCurrentTransactionId(TransactionId xid);
-extern bool CommandIdIsCurrentCommandId(CommandId cid);
 extern void CommandCounterIncrement(void);
 extern void StartTransactionCommand(void);
 extern void CommitTransactionCommand(void);
 extern void AbortCurrentTransaction(void);
 extern void BeginTransactionBlock(void);
 extern void EndTransactionBlock(void);
+extern bool IsSubTransaction(void);
 extern bool IsTransactionBlock(void);
 extern bool IsTransactionOrTransactionBlock(void);
 extern char TransactionBlockStatusCode(void);
@@ -151,6 +168,8 @@ extern void UnregisterEOXactCallback(EOXactCallback callback, void *arg);
 
 extern void RecordTransactionCommit(void);
 
+extern int	xactGetCommittedChildren(TransactionId **ptr, bool metoo);
+
 extern void XactPushRollback(void (*func) (void *), void *data);
 extern void XactPopRollback(void);
 
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 0e44e77446d..1c29ab07626 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -6,7 +6,7 @@
  * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/access/xlog.h,v 1.51 2004/05/29 22:48:22 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/access/xlog.h,v 1.52 2004/07/01 00:51:38 tgl Exp $
  */
 #ifndef XLOG_H
 #define XLOG_H
@@ -111,7 +111,7 @@ typedef struct XLogContRecord
 /*
  * Each page of XLOG file has a header like this:
  */
-#define XLOG_PAGE_MAGIC 0xD05A	/* can be used as WAL version indicator */
+#define XLOG_PAGE_MAGIC 0xD05B	/* can be used as WAL version indicator */
 
 typedef struct XLogPageHeaderData
 {