Change the bootstrap sequence so that toast tables for system catalogs are

created in the bootstrap phase proper, rather than added after-the-fact
by initdb.  This is cleaner than before because it allows us to retire the
undocumented ALTER TABLE ... CREATE TOAST TABLE command, but the real reason
I'm doing it is so that toast tables of shared catalogs will now have
predetermined OIDs.  This will allow a reasonably clean solution to the
problem of locking tables before we load their relcache entries, to appear
in a forthcoming patch.

src/backend/catalog/Makefile

@@ -2,7 +2,7 @@
 #
 # Makefile for backend/catalog
 #
-# $PostgreSQL: pgsql/src/backend/catalog/Makefile,v 1.59 2006/02/12 03:22:17 momjian Exp $
+# $PostgreSQL: pgsql/src/backend/catalog/Makefile,v 1.60 2006/07/31 01:16:36 tgl Exp $
 #
 #-------------------------------------------------------------------------
 
@@ -13,7 +13,7 @@ include $(top_builddir)/src/Makefile.global
 OBJS = catalog.o dependency.o heap.o index.o indexing.o namespace.o aclchk.o \
        pg_aggregate.o pg_constraint.o pg_conversion.o pg_depend.o \
        pg_largeobject.o pg_namespace.o pg_operator.o pg_proc.o pg_shdepend.o \
-       pg_type.o
+       pg_type.o toasting.o
 
 BKIFILES = postgres.bki postgres.description postgres.shdescription
 
@@ -24,7 +24,7 @@ SUBSYS.o: $(OBJS)
 
 # Note: there are some undocumented dependencies on the ordering in which
 # the catalog header files are assembled into postgres.bki.  In particular,
-# indexing.h had better be last.
+# indexing.h had better be last, and toasting.h just before it.
 
 POSTGRES_BKI_SRCS := $(addprefix $(top_srcdir)/src/include/catalog/,\
 	pg_proc.h pg_type.h pg_attribute.h pg_class.h pg_autovacuum.h \
@@ -35,7 +35,7 @@ POSTGRES_BKI_SRCS := $(addprefix $(top_srcdir)/src/include/catalog/,\
 	pg_namespace.h pg_conversion.h pg_depend.h \
 	pg_database.h pg_tablespace.h pg_pltemplate.h \
 	pg_authid.h pg_auth_members.h pg_shdepend.h pg_shdescription.h \
-	indexing.h \
+	toasting.h indexing.h \
     )
 
 pg_includes := $(sort -I$(top_srcdir)/src/include -I$(top_builddir)/src/include)

src/backend/catalog/README

@@ -1,4 +1,4 @@
-$PostgreSQL: pgsql/src/backend/catalog/README,v 1.9 2005/04/14 01:38:15 tgl Exp $
+$PostgreSQL: pgsql/src/backend/catalog/README,v 1.10 2006/07/31 01:16:36 tgl Exp $
 
 This directory contains .c files that manipulate the system catalogs;
 src/include/catalog contains the .h files that define the structure
@@ -73,8 +73,9 @@ heap_create_with_catalog process, because it needs these tables to exist
 already.  The list of files this currently includes is:
 	pg_proc.h pg_type.h pg_attribute.h pg_class.h
 Also, indexing.h must be last, since the indexes can't be created until all
-the tables are in place.  There are reputedly some other order dependencies
-in the .bki list, too.
+the tables are in place, and toasting.h should probably be next-to-last
+(or at least after all the tables that need toast tables).  There are
+reputedly some other order dependencies in the .bki list, too.
 
 -----------------------------------------------------------------
 

src/backend/catalog/genbki.sh

@@ -11,7 +11,7 @@
 #
 #
 # IDENTIFICATION
-#    $PostgreSQL: pgsql/src/backend/catalog/genbki.sh,v 1.39 2006/03/05 15:58:22 momjian Exp $
+#    $PostgreSQL: pgsql/src/backend/catalog/genbki.sh,v 1.40 2006/07/31 01:16:36 tgl Exp $
 #
 # NOTES
 #    non-essential whitespace is removed from the generated file.
@@ -276,6 +276,28 @@ comment_level > 0 { next; }
 	print "declare unique index " iname " " oid " " data
 }
 
+/^DECLARE_TOAST\(/ {
+# ----
+#  end any prior catalog data insertions before starting a define toast
+# ----
+	if (reln_open == 1) {
+		print "close " catalog;
+		reln_open = 0;
+	}
+
+	data = substr($0, 15, length($0) - 15);
+	pos = index(data, ",");
+	tname = substr(data, 1, pos-1);
+	data = substr(data, pos+1, length(data)-pos);
+	pos = index(data, ",");
+	toastoid = substr(data, 1, pos-1);
+	data = substr(data, pos+1, length(data)-pos);
+	# previous commands already removed the trailing );
+	indexoid = data;
+
+	print "declare toast " toastoid " " indexoid " on " tname
+}
+
 /^BUILD_INDICES/	{ print "build indices"; }
 	
 # ----------------

src/backend/catalog/heap.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/catalog/heap.c,v 1.309 2006/07/14 14:52:17 momjian Exp $
+ *	  $PostgreSQL: pgsql/src/backend/catalog/heap.c,v 1.310 2006/07/31 01:16:36 tgl Exp $
  *
  *
  * INTERFACE ROUTINES
@@ -2009,8 +2009,8 @@ RelationTruncateIndexes(Oid heapId)
 		RelationTruncate(currentIndex, 0);
 
 		/* Initialize the index and rebuild */
-		/* Note: we do not need to re-establish pkey or toast settings */
-		index_build(heapRelation, currentIndex, indexInfo, false, false);
+		/* Note: we do not need to re-establish pkey setting */
+		index_build(heapRelation, currentIndex, indexInfo, false);
 
 		/* We're done with this index */
 		index_close(currentIndex);

src/backend/catalog/index.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/catalog/index.c,v 1.270 2006/07/30 02:07:18 alvherre Exp $
+ *	  $PostgreSQL: pgsql/src/backend/catalog/index.c,v 1.271 2006/07/31 01:16:36 tgl Exp $
  *
  *
  * INTERFACE ROUTINES
@@ -424,7 +424,6 @@ UpdateIndexRelation(Oid indexoid,
  * classObjectId: array of index opclass OIDs, one per index column
  * reloptions: AM-specific options
  * isprimary: index is a PRIMARY KEY
- * istoast: index is a toast table's index
  * isconstraint: index is owned by a PRIMARY KEY or UNIQUE constraint
  * allow_system_table_mods: allow table to be a system catalog
  * skip_build: true to skip the index_build() step for the moment; caller
@@ -442,7 +441,6 @@ index_create(Oid heapRelationId,
 			 Oid *classObjectId,
 			 Datum reloptions,
 			 bool isprimary,
-			 bool istoast,
 			 bool isconstraint,
 			 bool allow_system_table_mods,
 			 bool skip_build)
@@ -747,8 +745,7 @@ index_create(Oid heapRelationId,
 	}
 	else
 	{
-		index_build(heapRelation, indexRelation, indexInfo,
-					isprimary, istoast);
+		index_build(heapRelation, indexRelation, indexInfo, isprimary);
 	}
 
 	/*
@@ -1241,8 +1238,8 @@ setNewRelfilenode(Relation relation)
  * entries of the index and heap relation as needed, using statistics
  * returned by ambuild as well as data passed by the caller.
  *
- * Note: when reindexing an existing index, isprimary and istoast can be
- * false; the index is already properly marked and need not be re-marked.
+ * Note: when reindexing an existing index, isprimary can be false;
+ * the index is already properly marked and need not be re-marked.
  *
  * Note: before Postgres 8.2, the passed-in heap and index Relations
  * were automatically closed by this routine.  This is no longer the case.
@@ -1252,8 +1249,7 @@ void
 index_build(Relation heapRelation,
 			Relation indexRelation,
 			IndexInfo *indexInfo,
-			bool isprimary,
-			bool istoast)
+			bool isprimary)
 {
 	RegProcedure procedure;
 	IndexBuildResult *stats;
@@ -1283,7 +1279,8 @@ index_build(Relation heapRelation,
 	index_update_stats(heapRelation,
 					   true,
 					   isprimary,
-					   istoast ? RelationGetRelid(indexRelation) : InvalidOid,
+					   (heapRelation->rd_rel->relkind == RELKIND_TOASTVALUE) ?
+					   RelationGetRelid(indexRelation) : InvalidOid,
 					   stats->heap_tuples);
 
 	index_update_stats(indexRelation,
@@ -1618,8 +1615,8 @@ reindex_index(Oid indexId)
 		}
 
 		/* Initialize the index and rebuild */
-		/* Note: we do not need to re-establish pkey or toast settings */
-		index_build(heapRelation, iRel, indexInfo, false, false);
+		/* Note: we do not need to re-establish pkey setting */
+		index_build(heapRelation, iRel, indexInfo, false);
 	}
 	PG_CATCH();
 	{

src/backend/catalog/toasting.c

@@ -0,0 +1,339 @@
+/*-------------------------------------------------------------------------
+ *
+ * toasting.c
+ *	  This file contains routines to support creation of toast tables
+ *
+ *
+ * Portions Copyright (c) 1996-2006, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  $PostgreSQL: pgsql/src/backend/catalog/toasting.c,v 1.1 2006/07/31 01:16:37 tgl Exp $
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "access/heapam.h"
+#include "access/tuptoaster.h"
+#include "access/xact.h"
+#include "catalog/dependency.h"
+#include "catalog/heap.h"
+#include "catalog/index.h"
+#include "catalog/indexing.h"
+#include "catalog/pg_namespace.h"
+#include "catalog/pg_opclass.h"
+#include "catalog/pg_type.h"
+#include "catalog/toasting.h"
+#include "miscadmin.h"
+#include "nodes/makefuncs.h"
+#include "utils/builtins.h"
+#include "utils/syscache.h"
+
+
+static bool create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid);
+static bool needs_toast_table(Relation rel);
+
+
+/*
+ * AlterTableCreateToastTable
+ *		If the table needs a toast table, and doesn't already have one,
+ *		then create a toast table for it.
+ *
+ * We expect the caller to have verified that the relation is a table and have
+ * already done any necessary permission checks.  Callers expect this function
+ * to end with CommandCounterIncrement if it makes any changes.
+ */
+void
+AlterTableCreateToastTable(Oid relOid)
+{
+	Relation	rel;
+
+	/*
+	 * Grab an exclusive lock on the target table, which we will NOT release
+	 * until end of transaction.  (This is probably redundant in all present
+	 * uses...)
+	 */
+	rel = heap_open(relOid, AccessExclusiveLock);
+
+	/* create_toast_table does all the work */
+	(void) create_toast_table(rel, InvalidOid, InvalidOid);
+
+	heap_close(rel, NoLock);
+}
+
+/*
+ * Create a toast table during bootstrap
+ *
+ * Here we need to prespecify the OIDs of the toast table and its index
+ */
+void
+BootstrapToastTable(char *relName, Oid toastOid, Oid toastIndexOid)
+{
+	Relation rel;
+
+	rel = heap_openrv(makeRangeVar(NULL, relName), AccessExclusiveLock);
+
+	/* Note: during bootstrap may see uncataloged relation */
+	if (rel->rd_rel->relkind != RELKIND_RELATION &&
+		rel->rd_rel->relkind != RELKIND_UNCATALOGED)
+		ereport(ERROR,
+				(errcode(ERRCODE_WRONG_OBJECT_TYPE),
+				 errmsg("\"%s\" is not a table",
+						relName)));
+
+	/* create_toast_table does all the work */
+	if (!create_toast_table(rel, toastOid, toastIndexOid))
+		elog(ERROR, "\"%s\" does not require a toast table",
+			 relName);
+
+	heap_close(rel, NoLock);
+}
+
+
+/*
+ * create_toast_table --- internal workhorse
+ *
+ * rel is already opened and exclusive-locked
+ * toastOid and toastIndexOid are normally InvalidOid, but during
+ * bootstrap they can be nonzero to specify hand-assigned OIDs
+ */
+static bool
+create_toast_table(Relation rel, Oid toastOid, Oid toastIndexOid)
+{
+	Oid			relOid = RelationGetRelid(rel);
+	HeapTuple	reltup;
+	TupleDesc	tupdesc;
+	bool		shared_relation;
+	Relation	class_rel;
+	Oid			toast_relid;
+	Oid			toast_idxid;
+	char		toast_relname[NAMEDATALEN];
+	char		toast_idxname[NAMEDATALEN];
+	IndexInfo  *indexInfo;
+	Oid			classObjectId[2];
+	ObjectAddress baseobject,
+				toastobject;
+
+	/*
+	 * Toast table is shared if and only if its parent is.
+	 *
+	 * We cannot allow toasting a shared relation after initdb (because
+	 * there's no way to mark it toasted in other databases' pg_class).
+	 */
+	shared_relation = rel->rd_rel->relisshared;
+	if (shared_relation && !IsBootstrapProcessingMode())
+		ereport(ERROR,
+				(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+				 errmsg("shared tables cannot be toasted after initdb")));
+
+	/*
+	 * Is it already toasted?
+	 */
+	if (rel->rd_rel->reltoastrelid != InvalidOid)
+		return false;
+
+	/*
+	 * Check to see whether the table actually needs a TOAST table.
+	 */
+	if (!needs_toast_table(rel))
+		return false;
+
+	/*
+	 * Create the toast table and its index
+	 */
+	snprintf(toast_relname, sizeof(toast_relname),
+			 "pg_toast_%u", relOid);
+	snprintf(toast_idxname, sizeof(toast_idxname),
+			 "pg_toast_%u_index", relOid);
+
+	/* this is pretty painful...  need a tuple descriptor */
+	tupdesc = CreateTemplateTupleDesc(3, false);
+	TupleDescInitEntry(tupdesc, (AttrNumber) 1,
+					   "chunk_id",
+					   OIDOID,
+					   -1, 0);
+	TupleDescInitEntry(tupdesc, (AttrNumber) 2,
+					   "chunk_seq",
+					   INT4OID,
+					   -1, 0);
+	TupleDescInitEntry(tupdesc, (AttrNumber) 3,
+					   "chunk_data",
+					   BYTEAOID,
+					   -1, 0);
+
+	/*
+	 * Ensure that the toast table doesn't itself get toasted, or we'll be
+	 * toast :-(.  This is essential for chunk_data because type bytea is
+	 * toastable; hit the other two just to be sure.
+	 */
+	tupdesc->attrs[0]->attstorage = 'p';
+	tupdesc->attrs[1]->attstorage = 'p';
+	tupdesc->attrs[2]->attstorage = 'p';
+
+	/*
+	 * Note: the toast relation is placed in the regular pg_toast namespace
+	 * even if its master relation is a temp table.  There cannot be any
+	 * naming collision, and the toast rel will be destroyed when its master
+	 * is, so there's no need to handle the toast rel as temp.
+	 *
+	 * XXX would it make sense to apply the master's reloptions to the toast
+	 * table?
+	 */
+	toast_relid = heap_create_with_catalog(toast_relname,
+										   PG_TOAST_NAMESPACE,
+										   rel->rd_rel->reltablespace,
+										   toastOid,
+										   rel->rd_rel->relowner,
+										   tupdesc,
+										   RELKIND_TOASTVALUE,
+										   shared_relation,
+										   true,
+										   0,
+										   ONCOMMIT_NOOP,
+										   (Datum) 0,
+										   true);
+
+	/* make the toast relation visible, else index creation will fail */
+	CommandCounterIncrement();
+
+	/*
+	 * Create unique index on chunk_id, chunk_seq.
+	 *
+	 * NOTE: the normal TOAST access routines could actually function with a
+	 * single-column index on chunk_id only. However, the slice access
+	 * routines use both columns for faster access to an individual chunk. In
+	 * addition, we want it to be unique as a check against the possibility of
+	 * duplicate TOAST chunk OIDs. The index might also be a little more
+	 * efficient this way, since btree isn't all that happy with large numbers
+	 * of equal keys.
+	 */
+
+	indexInfo = makeNode(IndexInfo);
+	indexInfo->ii_NumIndexAttrs = 2;
+	indexInfo->ii_KeyAttrNumbers[0] = 1;
+	indexInfo->ii_KeyAttrNumbers[1] = 2;
+	indexInfo->ii_Expressions = NIL;
+	indexInfo->ii_ExpressionsState = NIL;
+	indexInfo->ii_Predicate = NIL;
+	indexInfo->ii_PredicateState = NIL;
+	indexInfo->ii_Unique = true;
+
+	classObjectId[0] = OID_BTREE_OPS_OID;
+	classObjectId[1] = INT4_BTREE_OPS_OID;
+
+	toast_idxid = index_create(toast_relid, toast_idxname, toastIndexOid,
+							   indexInfo,
+							   BTREE_AM_OID,
+							   rel->rd_rel->reltablespace,
+							   classObjectId, (Datum) 0,
+							   true, false, true, false);
+
+	/*
+	 * Store the toast table's OID in the parent relation's pg_class row
+	 */
+	class_rel = heap_open(RelationRelationId, RowExclusiveLock);
+
+	reltup = SearchSysCacheCopy(RELOID,
+								ObjectIdGetDatum(relOid),
+								0, 0, 0);
+	if (!HeapTupleIsValid(reltup))
+		elog(ERROR, "cache lookup failed for relation %u", relOid);
+
+	((Form_pg_class) GETSTRUCT(reltup))->reltoastrelid = toast_relid;
+
+	if (!IsBootstrapProcessingMode())
+	{
+		/* normal case, use a transactional update */
+		simple_heap_update(class_rel, &reltup->t_self, reltup);
+
+		/* Keep catalog indexes current */
+		CatalogUpdateIndexes(class_rel, reltup);
+	}
+	else
+	{
+		/* While bootstrapping, we cannot UPDATE, so overwrite in-place */
+		heap_inplace_update(class_rel, reltup);
+	}
+
+	heap_freetuple(reltup);
+
+	heap_close(class_rel, RowExclusiveLock);
+
+	/*
+	 * Register dependency from the toast table to the master, so that the
+	 * toast table will be deleted if the master is.  Skip this in bootstrap
+	 * mode.
+	 */
+	if (!IsBootstrapProcessingMode())
+	{
+		baseobject.classId = RelationRelationId;
+		baseobject.objectId = relOid;
+		baseobject.objectSubId = 0;
+		toastobject.classId = RelationRelationId;
+		toastobject.objectId = toast_relid;
+		toastobject.objectSubId = 0;
+
+		recordDependencyOn(&toastobject, &baseobject, DEPENDENCY_INTERNAL);
+	}
+
+	/*
+	 * Make changes visible
+	 */
+	CommandCounterIncrement();
+
+	return true;
+}
+
+/*
+ * Check to see whether the table needs a TOAST table.	It does only if
+ * (1) there are any toastable attributes, and (2) the maximum length
+ * of a tuple could exceed TOAST_TUPLE_THRESHOLD.  (We don't want to
+ * create a toast table for something like "f1 varchar(20)".)
+ */
+static bool
+needs_toast_table(Relation rel)
+{
+	int32		data_length = 0;
+	bool		maxlength_unknown = false;
+	bool		has_toastable_attrs = false;
+	TupleDesc	tupdesc;
+	Form_pg_attribute *att;
+	int32		tuple_length;
+	int			i;
+
+	tupdesc = rel->rd_att;
+	att = tupdesc->attrs;
+
+	for (i = 0; i < tupdesc->natts; i++)
+	{
+		if (att[i]->attisdropped)
+			continue;
+		data_length = att_align(data_length, att[i]->attalign);
+		if (att[i]->attlen > 0)
+		{
+			/* Fixed-length types are never toastable */
+			data_length += att[i]->attlen;
+		}
+		else
+		{
+			int32		maxlen = type_maximum_size(att[i]->atttypid,
+												   att[i]->atttypmod);
+
+			if (maxlen < 0)
+				maxlength_unknown = true;
+			else
+				data_length += maxlen;
+			if (att[i]->attstorage != 'p')
+				has_toastable_attrs = true;
+		}
+	}
+	if (!has_toastable_attrs)
+		return false;			/* nothing to toast? */
+	if (maxlength_unknown)
+		return true;			/* any unlimited-length attrs? */
+	tuple_length = MAXALIGN(offsetof(HeapTupleHeaderData, t_bits) +
+							BITMAPLEN(tupdesc->natts)) +
+		MAXALIGN(data_length);
+	return (tuple_length > TOAST_TUPLE_THRESHOLD);
+}