Repair problems with VACUUM destroying t_ctid chains too soon, and with

insufficient paranoia in code that follows t_ctid links.  (We must do both
because even with VACUUM doing it properly, the intermediate state with
a dangling t_ctid link is visible concurrently during lazy VACUUM, and
could be seen afterwards if either type of VACUUM crashes partway through.)
Also try to improve documentation about what's going on.  Patch is a bit
bulky because passing the XMAX information around required changing the
APIs of some low-level heapam.c routines, but it's not conceptually very
complicated.  Per trouble report from Teodor and subsequent analysis.
This needs to be back-patched, but I'll do that after 8.1 beta is out.

src/backend/commands/async.c

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/commands/async.c,v 1.123 2005/06/17 22:32:43 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/commands/async.c,v 1.124 2005/08/20 00:39:53 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -550,8 +550,9 @@ AtCommit_Notify(void)
 			}
 			else if (listener->notification == 0)
 			{
-				ItemPointerData ctid;
 				HTSU_Result		result;
+				ItemPointerData update_ctid;
+				TransactionId update_xmax;
 
 				rTuple = heap_modifytuple(lTuple, tdesc,
 										  value, nulls, repl);
@@ -573,7 +574,7 @@ AtCommit_Notify(void)
 				 * heap_update calls.
 				 */
 				result = heap_update(lRel, &lTuple->t_self, rTuple,
-									 &ctid,
+									 &update_ctid, &update_xmax,
 									 GetCurrentCommandId(), InvalidSnapshot,
 									 false /* no wait for commit */ );
 				switch (result)
@@ -585,7 +586,6 @@ AtCommit_Notify(void)
 
 					case HeapTupleMayBeUpdated:
 						/* done successfully */
-
 #ifdef NOT_USED					/* currently there are no indexes */
 						CatalogUpdateIndexes(lRel, rTuple);
 #endif