Assert that buffers are marked dirty before XLogRegisterBuffer().

Enforce the rule from transam/README in XLogRegisterBuffer(), and
update callers to follow the rule.

Hash indexes sometimes register clean pages as a part of the locking
protocol, so provide a REGBUF_NO_CHANGE flag to support that use.

Discussion: https://postgr.es/m/c84114f8-c7f1-5b57-f85a-3adc31e1a904@iki.fi
Reviewed-by: Heikki Linnakangas

doc/src/sgml/monitoring.sgml

@@ -1119,7 +1119,7 @@ SELECT a.pid, a.wait_event, w.description
   FROM pg_stat_activity a JOIN
        pg_wait_events w ON (a.wait_event_type = w.type AND
                             a.wait_event = w.name)
-  WHERE wait_event is NOT NULL and a.state = 'active';
+  WHERE a.wait_event is NOT NULL and a.state = 'active';
 -[ RECORD 1 ]------------------------------------------------------&zwsp;------------
 pid         | 686674
 wait_event  | WALInitSync

src/backend/access/gin/ginbtree.c

@@ -387,24 +387,22 @@ ginPlaceToPage(GinBtree btree, GinBtreeStack *stack,
 		START_CRIT_SECTION();
 
 		if (RelationNeedsWAL(btree->index) && !btree->isBuild)
-		{
 			XLogBeginInsert();
-			XLogRegisterBuffer(0, stack->buffer, REGBUF_STANDARD);
-			if (BufferIsValid(childbuf))
-				XLogRegisterBuffer(1, childbuf, REGBUF_STANDARD);
-		}
 
-		/* Perform the page update, and register any extra WAL data */
+		/*
+		 * Perform the page update, dirty and register stack->buffer, and
+		 * register any extra WAL data.
+		 */
 		btree->execPlaceToPage(btree, stack->buffer, stack,
 							   insertdata, updateblkno, ptp_workspace);
 
-		MarkBufferDirty(stack->buffer);
-
 		/* An insert to an internal page finishes the split of the child. */
 		if (BufferIsValid(childbuf))
 		{
 			GinPageGetOpaque(childpage)->flags &= ~GIN_INCOMPLETE_SPLIT;
 			MarkBufferDirty(childbuf);
+			if (RelationNeedsWAL(btree->index) && !btree->isBuild)
+				XLogRegisterBuffer(1, childbuf, REGBUF_STANDARD);
 		}
 
 		if (RelationNeedsWAL(btree->index) && !btree->isBuild)

src/backend/access/gin/gindatapage.c

@@ -721,9 +721,12 @@ dataExecPlaceToPageLeaf(GinBtree btree, Buffer buf, GinBtreeStack *stack,
 	/* Apply changes to page */
 	dataPlaceToPageLeafRecompress(buf, leaf);
 
+	MarkBufferDirty(buf);
+
 	/* If needed, register WAL data built by computeLeafRecompressWALData */
 	if (RelationNeedsWAL(btree->index) && !btree->isBuild)
 	{
+		XLogRegisterBuffer(0, buf, REGBUF_STANDARD);
 		XLogRegisterBufData(0, leaf->walinfo, leaf->walinfolen);
 	}
 }
@@ -1155,6 +1158,8 @@ dataExecPlaceToPageInternal(GinBtree btree, Buffer buf, GinBtreeStack *stack,
 	pitem = (PostingItem *) insertdata;
 	GinDataPageAddPostingItem(page, pitem, off);
 
+	MarkBufferDirty(buf);
+
 	if (RelationNeedsWAL(btree->index) && !btree->isBuild)
 	{
 		/*
@@ -1167,6 +1172,7 @@ dataExecPlaceToPageInternal(GinBtree btree, Buffer buf, GinBtreeStack *stack,
 		data.offset = off;
 		data.newitem = *pitem;
 
+		XLogRegisterBuffer(0, buf, REGBUF_STANDARD);
 		XLogRegisterBufData(0, (char *) &data,
 							sizeof(ginxlogInsertDataInternal));
 	}

src/backend/access/gin/ginentrypage.c

@@ -571,6 +571,8 @@ entryExecPlaceToPage(GinBtree btree, Buffer buf, GinBtreeStack *stack,
 		elog(ERROR, "failed to add item to index page in \"%s\"",
 			 RelationGetRelationName(btree->index));
 
+	MarkBufferDirty(buf);
+
 	if (RelationNeedsWAL(btree->index) && !btree->isBuild)
 	{
 		/*
@@ -583,6 +585,7 @@ entryExecPlaceToPage(GinBtree btree, Buffer buf, GinBtreeStack *stack,
 		data.isDelete = insertData->isDelete;
 		data.offset = off;
 
+		XLogRegisterBuffer(0, buf, REGBUF_STANDARD);
 		XLogRegisterBufData(0, (char *) &data,
 							offsetof(ginxlogInsertEntry, tuple));
 		XLogRegisterBufData(0, (char *) insertData->entry,

src/backend/access/gin/ginfast.c

@@ -397,6 +397,9 @@ ginHeapTupleFastInsert(GinState *ginstate, GinTupleCollector *collector)
 		}
 
 		Assert((ptr - collectordata) <= collector->sumsize);
+
+		MarkBufferDirty(buffer);
+
 		if (needWal)
 		{
 			XLogRegisterBuffer(1, buffer, REGBUF_STANDARD);
@@ -404,8 +407,6 @@ ginHeapTupleFastInsert(GinState *ginstate, GinTupleCollector *collector)
 		}
 
 		metadata->tailFreeSize = PageGetExactFreeSpace(page);
-
-		MarkBufferDirty(buffer);
 	}
 
 	/*

src/backend/access/hash/hash.c

@@ -824,11 +824,16 @@ hashbucketcleanup(Relation rel, Bucket cur_bucket, Buffer bucket_buf,
 				XLogRegisterData((char *) &xlrec, SizeOfHashDelete);
 
 				/*
-				 * bucket buffer needs to be registered to ensure that we can
+				 * bucket buffer was not changed, but still needs to be
-				 * acquire a cleanup lock on it during replay.
+				 * registered to ensure that we can acquire a cleanup lock on
+				 * it during replay.
 				 */
 				if (!xlrec.is_primary_bucket_page)
-					XLogRegisterBuffer(0, bucket_buf, REGBUF_STANDARD | REGBUF_NO_IMAGE);
+				{
+					uint8		flags = REGBUF_STANDARD | REGBUF_NO_IMAGE | REGBUF_NO_CHANGE;
+
+					XLogRegisterBuffer(0, bucket_buf, flags);
+				}
 
 				XLogRegisterBuffer(1, buf, REGBUF_STANDARD);
 				XLogRegisterBufData(1, (char *) deletable,

src/backend/access/hash/hashovfl.c

@@ -658,11 +658,15 @@ _hash_freeovflpage(Relation rel, Buffer bucketbuf, Buffer ovflbuf,
 		XLogRegisterData((char *) &xlrec, SizeOfHashSqueezePage);
 
 		/*
-		 * bucket buffer needs to be registered to ensure that we can acquire
+		 * bucket buffer was not changed, but still needs to be registered to
-		 * a cleanup lock on it during replay.
+		 * ensure that we can acquire a cleanup lock on it during replay.
 		 */
 		if (!xlrec.is_prim_bucket_same_wrt)
-			XLogRegisterBuffer(0, bucketbuf, REGBUF_STANDARD | REGBUF_NO_IMAGE);
+		{
+			uint8		flags = REGBUF_STANDARD | REGBUF_NO_IMAGE | REGBUF_NO_CHANGE;
+
+			XLogRegisterBuffer(0, bucketbuf, flags);
+		}
 
 		XLogRegisterBuffer(1, wbuf, REGBUF_STANDARD);
 		if (xlrec.ntups > 0)
@@ -960,11 +964,16 @@ readpage:
 						XLogRegisterData((char *) &xlrec, SizeOfHashMovePageContents);
 
 						/*
-						 * bucket buffer needs to be registered to ensure that
+						 * bucket buffer was not changed, but still needs to
-						 * we can acquire a cleanup lock on it during replay.
+						 * be registered to ensure that we can acquire a
+						 * cleanup lock on it during replay.
 						 */
 						if (!xlrec.is_prim_bucket_same_wrt)
-							XLogRegisterBuffer(0, bucket_buf, REGBUF_STANDARD | REGBUF_NO_IMAGE);
+						{
+							int			flags = REGBUF_STANDARD | REGBUF_NO_IMAGE | REGBUF_NO_CHANGE;
+
+							XLogRegisterBuffer(0, bucket_buf, flags);
+						}
 
 						XLogRegisterBuffer(1, wbuf, REGBUF_STANDARD);
 						XLogRegisterBufData(1, (char *) itup_offsets,

src/backend/access/transam/xlog.c

@@ -8579,9 +8579,22 @@ do_pg_backup_start(const char *backupidstr, bool fast, List **tablespaces,
 			char	   *relpath = NULL;
 			char	   *s;
 			PGFileType	de_type;
+			char	   *badp;
+			Oid			tsoid;
 
-			/* Skip anything that doesn't look like a tablespace */
+			/*
-			if (strspn(de->d_name, "0123456789") != strlen(de->d_name))
+			 * Try to parse the directory name as an unsigned integer.
+			 *
+			 * Tablespace directories should be positive integers that can be
+			 * represented in 32 bits, with no leading zeroes or trailing
+			 * garbage. If we come across a name that doesn't meet those
+			 * criteria, skip it.
+			 */
+			if (de->d_name[0] < '1' || de->d_name[1] > '9')
+				continue;
+			errno = 0;
+			tsoid = strtoul(de->d_name, &badp, 10);
+			if (*badp != '\0' || errno == EINVAL || errno == ERANGE)
 				continue;
 
 			snprintf(fullpath, sizeof(fullpath), "pg_tblspc/%s", de->d_name);
@@ -8656,7 +8669,7 @@ do_pg_backup_start(const char *backupidstr, bool fast, List **tablespaces,
 			}
 
 			ti = palloc(sizeof(tablespaceinfo));
-			ti->oid = pstrdup(de->d_name);
+			ti->oid = tsoid;
 			ti->path = pstrdup(linkpath);
 			ti->rpath = relpath;
 			ti->size = -1;

src/backend/access/transam/xloginsert.c

@@ -248,6 +248,20 @@ XLogRegisterBuffer(uint8 block_id, Buffer buffer, uint8 flags)
 	Assert(!((flags & REGBUF_FORCE_IMAGE) && (flags & (REGBUF_NO_IMAGE))));
 	Assert(begininsert_called);
 
+	/*
+	 * Ordinarily, buffer should be exclusive-locked and marked dirty before
+	 * we get here, otherwise we could end up violating one of the rules in
+	 * access/transam/README.
+	 *
+	 * Some callers intentionally register a clean page and never update that
+	 * page's LSN; in that case they can pass the flag REGBUF_NO_CHANGE to
+	 * bypass these checks.
+	 */
+#ifdef USE_ASSERT_CHECKING
+	if (!(flags & REGBUF_NO_CHANGE))
+		Assert(BufferIsExclusiveLocked(buffer) && BufferIsDirty(buffer));
+#endif
+
 	if (block_id >= max_registered_block_id)
 	{
 		if (block_id >= max_registered_buffers)
@@ -1313,8 +1327,8 @@ log_newpage_range(Relation rel, ForkNumber forknum,
 		START_CRIT_SECTION();
 		for (i = 0; i < nbufs; i++)
 		{
-			XLogRegisterBuffer(i, bufpack[i], flags);
 			MarkBufferDirty(bufpack[i]);
+			XLogRegisterBuffer(i, bufpack[i], flags);
 		}
 
 		recptr = XLogInsert(RM_XLOG_ID, XLOG_FPI);

src/backend/access/transam/xlogrecovery.c

@@ -678,7 +678,7 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr,
 				tablespaceinfo *ti = lfirst(lc);
 				char	   *linkloc;
 
-				linkloc = psprintf("pg_tblspc/%s", ti->oid);
+				linkloc = psprintf("pg_tblspc/%u", ti->oid);
 
 				/*
 				 * Remove the existing symlink if any and Create the symlink
@@ -692,7 +692,6 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr,
 							 errmsg("could not create symbolic link \"%s\": %m",
 									linkloc)));
 
-				pfree(ti->oid);
 				pfree(ti->path);
 				pfree(ti);
 			}
@@ -1341,6 +1340,8 @@ read_tablespace_map(List **tablespaces)
 	{
 		if (!was_backslash && (ch == '\n' || ch == '\r'))
 		{
+			char	   *endp;
+
 			if (i == 0)
 				continue;		/* \r immediately followed by \n */
 
@@ -1360,7 +1361,12 @@ read_tablespace_map(List **tablespaces)
 			str[n++] = '\0';
 
 			ti = palloc0(sizeof(tablespaceinfo));
-			ti->oid = pstrdup(str);
+			errno = 0;
+			ti->oid = strtoul(str, &endp, 10);
+			if (*endp != '\0' || errno == EINVAL || errno == ERANGE)
+				ereport(FATAL,
+						(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+						 errmsg("invalid data in file \"%s\"", TABLESPACE_MAP)));
 			ti->path = pstrdup(str + n);
 			*tablespaces = lappend(*tablespaces, ti);
 

src/backend/backup/backup_manifest.c

@@ -97,7 +97,7 @@ FreeBackupManifest(backup_manifest_info *manifest)
  * Add an entry to the backup manifest for a file.
  */
 void
-AddFileToBackupManifest(backup_manifest_info *manifest, const char *spcoid,
+AddFileToBackupManifest(backup_manifest_info *manifest, Oid spcoid,
 						const char *pathname, size_t size, pg_time_t mtime,
 						pg_checksum_context *checksum_ctx)
 {
@@ -114,9 +114,9 @@ AddFileToBackupManifest(backup_manifest_info *manifest, const char *spcoid,
 	 * pathname relative to the data directory (ignoring the intermediate
 	 * symlink traversal).
 	 */
-	if (spcoid != NULL)
+	if (OidIsValid(spcoid))
 	{
-		snprintf(pathbuf, sizeof(pathbuf), "pg_tblspc/%s/%s", spcoid,
+		snprintf(pathbuf, sizeof(pathbuf), "pg_tblspc/%u/%s", spcoid,
 				 pathname);
 		pathname = pathbuf;
 	}

src/backend/backup/basebackup.c

@@ -75,14 +75,15 @@ typedef struct
 	pg_checksum_type manifest_checksum_type;
 } basebackup_options;
 
-static int64 sendTablespace(bbsink *sink, char *path, char *spcoid, bool sizeonly,
+static int64 sendTablespace(bbsink *sink, char *path, Oid spcoid, bool sizeonly,
 							struct backup_manifest_info *manifest);
 static int64 sendDir(bbsink *sink, const char *path, int basepathlen, bool sizeonly,
 					 List *tablespaces, bool sendtblspclinks,
-					 backup_manifest_info *manifest, const char *spcoid);
+					 backup_manifest_info *manifest, Oid spcoid);
 static bool sendFile(bbsink *sink, const char *readfilename, const char *tarfilename,
-					 struct stat *statbuf, bool missing_ok, Oid dboid,
+					 struct stat *statbuf, bool missing_ok,
-					 backup_manifest_info *manifest, const char *spcoid);
+					 Oid dboid, Oid spcoid,
+					 backup_manifest_info *manifest);
 static off_t read_file_data_into_buffer(bbsink *sink,
 										const char *readfilename, int fd,
 										off_t offset, size_t length,
@@ -305,7 +306,7 @@ perform_base_backup(basebackup_options *opt, bbsink *sink)
 
 				if (tmp->path == NULL)
 					tmp->size = sendDir(sink, ".", 1, true, state.tablespaces,
-										true, NULL, NULL);
+										true, NULL, InvalidOid);
 				else
 					tmp->size = sendTablespace(sink, tmp->path, tmp->oid, true,
 											   NULL);
@@ -346,7 +347,7 @@ perform_base_backup(basebackup_options *opt, bbsink *sink)
 
 				/* Then the bulk of the files... */
 				sendDir(sink, ".", 1, false, state.tablespaces,
-						sendtblspclinks, &manifest, NULL);
+						sendtblspclinks, &manifest, InvalidOid);
 
 				/* ... and pg_control after everything else. */
 				if (lstat(XLOG_CONTROL_FILE, &statbuf) != 0)
@@ -355,11 +356,11 @@ perform_base_backup(basebackup_options *opt, bbsink *sink)
 							 errmsg("could not stat file \"%s\": %m",
 									XLOG_CONTROL_FILE)));
 				sendFile(sink, XLOG_CONTROL_FILE, XLOG_CONTROL_FILE, &statbuf,
-						 false, InvalidOid, &manifest, NULL);
+						 false, InvalidOid, InvalidOid, &manifest);
 			}
 			else
 			{
-				char	   *archive_name = psprintf("%s.tar", ti->oid);
+				char	   *archive_name = psprintf("%u.tar", ti->oid);
 
 				bbsink_begin_archive(sink, archive_name);
 
@@ -623,8 +624,8 @@ perform_base_backup(basebackup_options *opt, bbsink *sink)
 						(errcode_for_file_access(),
 						 errmsg("could not stat file \"%s\": %m", pathbuf)));
 
-			sendFile(sink, pathbuf, pathbuf, &statbuf, false, InvalidOid,
+			sendFile(sink, pathbuf, pathbuf, &statbuf, false,
-					 &manifest, NULL);
+					 InvalidOid, InvalidOid, &manifest);
 
 			/* unconditionally mark file as archived */
 			StatusFilePath(pathbuf, fname, ".done");
@@ -1087,7 +1088,7 @@ sendFileWithContent(bbsink *sink, const char *filename, const char *content,
 
 	_tarWritePadding(sink, len);
 
-	AddFileToBackupManifest(manifest, NULL, filename, len,
+	AddFileToBackupManifest(manifest, InvalidOid, filename, len,
 							(pg_time_t) statbuf.st_mtime, &checksum_ctx);
 }
 
@@ -1099,7 +1100,7 @@ sendFileWithContent(bbsink *sink, const char *filename, const char *content,
  * Only used to send auxiliary tablespaces, not PGDATA.
  */
 static int64
-sendTablespace(bbsink *sink, char *path, char *spcoid, bool sizeonly,
+sendTablespace(bbsink *sink, char *path, Oid spcoid, bool sizeonly,
 			   backup_manifest_info *manifest)
 {
 	int64		size;
@@ -1154,7 +1155,7 @@ sendTablespace(bbsink *sink, char *path, char *spcoid, bool sizeonly,
 static int64
 sendDir(bbsink *sink, const char *path, int basepathlen, bool sizeonly,
 		List *tablespaces, bool sendtblspclinks, backup_manifest_info *manifest,
-		const char *spcoid)
+		Oid spcoid)
 {
 	DIR		   *dir;
 	struct dirent *de;
@@ -1197,9 +1198,9 @@ sendDir(bbsink *sink, const char *path, int basepathlen, bool sizeonly,
 	{
 		int			excludeIdx;
 		bool		excludeFound;
-		ForkNumber	relForkNum; /* Type of fork if file is a relation */
+		RelFileNumber relNumber;
-		int			relnumchars;	/* Chars in filename that are the
+		ForkNumber	relForkNum;
-									 * relnumber */
+		unsigned	segno;
 
 		/* Skip special stuff */
 		if (strcmp(de->d_name, ".") == 0 || strcmp(de->d_name, "..") == 0)
@@ -1249,23 +1250,20 @@ sendDir(bbsink *sink, const char *path, int basepathlen, bool sizeonly,
 
 		/* Exclude all forks for unlogged tables except the init fork */
 		if (isDbDir &&
-			parse_filename_for_nontemp_relation(de->d_name, &relnumchars,
+			parse_filename_for_nontemp_relation(de->d_name, &relNumber,
-												&relForkNum))
+												&relForkNum, &segno))
 		{
 			/* Never exclude init forks */
 			if (relForkNum != INIT_FORKNUM)
 			{
 				char		initForkFile[MAXPGPATH];
-				char		relNumber[OIDCHARS + 1];
 
 				/*
 				 * If any other type of fork, check if there is an init fork
 				 * with the same RelFileNumber. If so, the file can be
 				 * excluded.
 				 */
-				memcpy(relNumber, de->d_name, relnumchars);
+				snprintf(initForkFile, sizeof(initForkFile), "%s/%u_init",
-				relNumber[relnumchars] = '\0';
-				snprintf(initForkFile, sizeof(initForkFile), "%s/%s_init",
 						 path, relNumber);
 
 				if (lstat(initForkFile, &statbuf) == 0)
@@ -1419,8 +1417,8 @@ sendDir(bbsink *sink, const char *path, int basepathlen, bool sizeonly,
 
 			if (!sizeonly)
 				sent = sendFile(sink, pathbuf, pathbuf + basepathlen + 1, &statbuf,
-								true, isDbDir ? atooid(lastDir + 1) : InvalidOid,
+								true, isDbDir ? atooid(lastDir + 1) : InvalidOid, spcoid,
-								manifest, spcoid);
+								manifest);
 
 			if (sent || sizeonly)
 			{
@@ -1489,8 +1487,8 @@ is_checksummed_file(const char *fullpath, const char *filename)
  */
 static bool
 sendFile(bbsink *sink, const char *readfilename, const char *tarfilename,
-		 struct stat *statbuf, bool missing_ok, Oid dboid,
+		 struct stat *statbuf, bool missing_ok, Oid dboid, Oid spcoid,
-		 backup_manifest_info *manifest, const char *spcoid)
+		 backup_manifest_info *manifest)
 {
 	int			fd;
 	BlockNumber blkno = 0;

src/backend/backup/basebackup_copy.c

@@ -407,7 +407,7 @@ SendTablespaceList(List *tablespaces)
 		}
 		else
 		{
-			values[0] = ObjectIdGetDatum(strtoul(ti->oid, NULL, 10));
+			values[0] = ObjectIdGetDatum(ti->oid);
 			values[1] = CStringGetTextDatum(ti->path);
 		}
 		if (ti->size >= 0)

src/backend/storage/buffer/bufmgr.c

@@ -2098,6 +2098,65 @@ ExtendBufferedRelShared(BufferManagerRelation bmr,
 	return first_block;
 }
 
+/*
+ * BufferIsExclusiveLocked
+ *
+ *      Checks if buffer is exclusive-locked.
+ *
+ * Buffer must be pinned.
+ */
+bool
+BufferIsExclusiveLocked(Buffer buffer)
+{
+	BufferDesc *bufHdr;
+
+	if (BufferIsLocal(buffer))
+	{
+		int			bufid = -buffer - 1;
+
+		bufHdr = GetLocalBufferDescriptor(bufid);
+	}
+	else
+	{
+		bufHdr = GetBufferDescriptor(buffer - 1);
+	}
+
+	Assert(BufferIsPinned(buffer));
+	return LWLockHeldByMeInMode(BufferDescriptorGetContentLock(bufHdr),
+								LW_EXCLUSIVE);
+}
+
+/*
+ * BufferIsDirty
+ *
+ *		Checks if buffer is already dirty.
+ *
+ * Buffer must be pinned and exclusive-locked.  (Without an exclusive lock,
+ * the result may be stale before it's returned.)
+ */
+bool
+BufferIsDirty(Buffer buffer)
+{
+	BufferDesc *bufHdr;
+
+	if (BufferIsLocal(buffer))
+	{
+		int			bufid = -buffer - 1;
+
+		bufHdr = GetLocalBufferDescriptor(bufid);
+	}
+	else
+	{
+		bufHdr = GetBufferDescriptor(buffer - 1);
+	}
+
+	Assert(BufferIsPinned(buffer));
+	Assert(LWLockHeldByMeInMode(BufferDescriptorGetContentLock(bufHdr),
+								LW_EXCLUSIVE));
+
+	return pg_atomic_read_u32(&bufHdr->state) & BM_DIRTY;
+}
+
 /*
  * MarkBufferDirty
  *

src/backend/storage/file/reinit.c

@@ -31,7 +31,7 @@ static void ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname,
 
 typedef struct
 {
-	Oid			reloid;			/* hash key */
+	RelFileNumber relnumber;	/* hash key */
 } unlogged_relation_entry;
 
 /*
@@ -195,12 +195,13 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 		while ((de = ReadDir(dbspace_dir, dbspacedirname)) != NULL)
 		{
 			ForkNumber	forkNum;
-			int			relnumchars;
+			unsigned	segno;
 			unlogged_relation_entry ent;
 
 			/* Skip anything that doesn't look like a relation data file. */
-			if (!parse_filename_for_nontemp_relation(de->d_name, &relnumchars,
+			if (!parse_filename_for_nontemp_relation(de->d_name,
-													 &forkNum))
+													 &ent.relnumber,
+													 &forkNum, &segno))
 				continue;
 
 			/* Also skip it unless this is the init fork. */
@@ -208,10 +209,8 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 				continue;
 
 			/*
-			 * Put the OID portion of the name into the hash table, if it
+			 * Put the RelFileNumber into the hash table, if it isn't already.
-			 * isn't already.
 			 */
-			ent.reloid = atooid(de->d_name);
 			(void) hash_search(hash, &ent, HASH_ENTER, NULL);
 		}
 
@@ -235,12 +234,13 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 		while ((de = ReadDir(dbspace_dir, dbspacedirname)) != NULL)
 		{
 			ForkNumber	forkNum;
-			int			relnumchars;
+			unsigned	segno;
 			unlogged_relation_entry ent;
 
 			/* Skip anything that doesn't look like a relation data file. */
-			if (!parse_filename_for_nontemp_relation(de->d_name, &relnumchars,
+			if (!parse_filename_for_nontemp_relation(de->d_name,
-													 &forkNum))
+													 &ent.relnumber,
+													 &forkNum, &segno))
 				continue;
 
 			/* We never remove the init fork. */
@@ -251,7 +251,6 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 			 * See whether the OID portion of the name shows up in the hash
 			 * table.  If so, nuke it!
 			 */
-			ent.reloid = atooid(de->d_name);
 			if (hash_search(hash, &ent, HASH_FIND, NULL))
 			{
 				snprintf(rm_path, sizeof(rm_path), "%s/%s",
@@ -285,14 +284,14 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 		while ((de = ReadDir(dbspace_dir, dbspacedirname)) != NULL)
 		{
 			ForkNumber	forkNum;
-			int			relnumchars;
+			RelFileNumber relNumber;
-			char		relnumbuf[OIDCHARS + 1];
+			unsigned	segno;
 			char		srcpath[MAXPGPATH * 2];
 			char		dstpath[MAXPGPATH];
 
 			/* Skip anything that doesn't look like a relation data file. */
-			if (!parse_filename_for_nontemp_relation(de->d_name, &relnumchars,
+			if (!parse_filename_for_nontemp_relation(de->d_name, &relNumber,
-													 &forkNum))
+													 &forkNum, &segno))
 				continue;
 
 			/* Also skip it unless this is the init fork. */
@@ -304,11 +303,12 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 					 dbspacedirname, de->d_name);
 
 			/* Construct destination pathname. */
-			memcpy(relnumbuf, de->d_name, relnumchars);
+			if (segno == 0)
-			relnumbuf[relnumchars] = '\0';
+				snprintf(dstpath, sizeof(dstpath), "%s/%u",
-			snprintf(dstpath, sizeof(dstpath), "%s/%s%s",
+						 dbspacedirname, relNumber);
-					 dbspacedirname, relnumbuf, de->d_name + relnumchars + 1 +
+			else
-					 strlen(forkNames[INIT_FORKNUM]));
+				snprintf(dstpath, sizeof(dstpath), "%s/%u.%u",
+						 dbspacedirname, relNumber, segno);
 
 			/* OK, we're ready to perform the actual copy. */
 			elog(DEBUG2, "copying %s to %s", srcpath, dstpath);
@@ -327,14 +327,14 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 		dbspace_dir = AllocateDir(dbspacedirname);
 		while ((de = ReadDir(dbspace_dir, dbspacedirname)) != NULL)
 		{
+			RelFileNumber relNumber;
 			ForkNumber	forkNum;
-			int			relnumchars;
+			unsigned	segno;
-			char		relnumbuf[OIDCHARS + 1];
 			char		mainpath[MAXPGPATH];
 
 			/* Skip anything that doesn't look like a relation data file. */
-			if (!parse_filename_for_nontemp_relation(de->d_name, &relnumchars,
+			if (!parse_filename_for_nontemp_relation(de->d_name, &relNumber,
-													 &forkNum))
+													 &forkNum, &segno))
 				continue;
 
 			/* Also skip it unless this is the init fork. */
@@ -342,11 +342,12 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
 				continue;
 
 			/* Construct main fork pathname. */
-			memcpy(relnumbuf, de->d_name, relnumchars);
+			if (segno == 0)
-			relnumbuf[relnumchars] = '\0';
+				snprintf(mainpath, sizeof(mainpath), "%s/%u",
-			snprintf(mainpath, sizeof(mainpath), "%s/%s%s",
+						 dbspacedirname, relNumber);
-					 dbspacedirname, relnumbuf, de->d_name + relnumchars + 1 +
+			else
-					 strlen(forkNames[INIT_FORKNUM]));
+				snprintf(mainpath, sizeof(mainpath), "%s/%u.%u",
+						 dbspacedirname, relNumber, segno);
 
 			fsync_fname(mainpath, false);
 		}
@@ -371,52 +372,82 @@ ResetUnloggedRelationsInDbspaceDir(const char *dbspacedirname, int op)
  * This function returns true if the file appears to be in the correct format
  * for a non-temporary relation and false otherwise.
  *
- * NB: If this function returns true, the caller is entitled to assume that
+ * If it returns true, it sets *relnumber, *fork, and *segno to the values
- * *relnumchars has been set to a value no more than OIDCHARS, and thus
+ * extracted from the filename. If it returns false, these values are set to
- * that a buffer of OIDCHARS+1 characters is sufficient to hold the
+ * InvalidRelFileNumber, InvalidForkNumber, and 0, respectively.
- * RelFileNumber portion of the filename.  This is critical to protect against
- * a possible buffer overrun.
  */
 bool
-parse_filename_for_nontemp_relation(const char *name, int *relnumchars,
+parse_filename_for_nontemp_relation(const char *name, RelFileNumber *relnumber,
-									ForkNumber *fork)
+									ForkNumber *fork, unsigned *segno)
 {
-	int			pos;
+	unsigned long n,
+				s;
+	ForkNumber	f;
+	char	   *endp;
 
-	/* Look for a non-empty string of digits (that isn't too long). */
+	*relnumber = InvalidRelFileNumber;
-	for (pos = 0; isdigit((unsigned char) name[pos]); ++pos)
+	*fork = InvalidForkNumber;
-		;
+	*segno = 0;
-	if (pos == 0 || pos > OIDCHARS)
+
+	/*
+	 * Relation filenames should begin with a digit that is not a zero. By
+	 * rejecting cases involving leading zeroes, the caller can assume that
+	 * there's only one possible string of characters that could have produced
+	 * any given value for *relnumber.
+	 *
+	 * (To be clear, we don't expect files with names like 0017.3 to exist at
+	 * all -- but if 0017.3 does exist, it's a non-relation file, not part of
+	 * the main fork for relfilenode 17.)
+	 */
+	if (name[0] < '1' || name[0] > '9')
 		return false;
-	*relnumchars = pos;
+
+	/*
+	 * Parse the leading digit string. If the value is out of range, we
+	 * conclude that this isn't a relation file at all.
+	 */
+	errno = 0;
+	n = strtoul(name, &endp, 10);
+	if (errno || name == endp || n <= 0 || n > PG_UINT32_MAX)
+		return false;
+	name = endp;
 
 	/* Check for a fork name. */
-	if (name[pos] != '_')
+	if (*name != '_')
-		*fork = MAIN_FORKNUM;
+		f = MAIN_FORKNUM;
 	else
 	{
 		int			forkchar;
 
-		forkchar = forkname_chars(&name[pos + 1], fork);
+		forkchar = forkname_chars(name + 1, &f);
 		if (forkchar <= 0)
 			return false;
-		pos += forkchar + 1;
+		name += forkchar + 1;
 	}
 
 	/* Check for a segment number. */
-	if (name[pos] == '.')
+	if (*name != '.')
+		s = 0;
+	else
 	{
-		int			segchar;
+		/* Reject leading zeroes, just like we do for RelFileNumber. */
-
+		if (name[0] < '1' || name[0] > '9')
-		for (segchar = 1; isdigit((unsigned char) name[pos + segchar]); ++segchar)
-			;
-		if (segchar <= 1)
 			return false;
-		pos += segchar;
+
+		errno = 0;
+		s = strtoul(name + 1, &endp, 10);
+		if (errno || name + 1 == endp || s <= 0 || s > PG_UINT32_MAX)
+			return false;
+		name = endp;
 	}
 
 	/* Now we should be at the end. */
-	if (name[pos] != '\0')
+	if (*name != '\0')
 		return false;
+
+	/* Set out parameters and return. */
+	*relnumber = (RelFileNumber) n;
+	*fork = f;
+	*segno = (unsigned) s;
 	return true;
 }

src/backend/utils/activity/wait_event.c

@@ -56,9 +56,9 @@ uint32	   *my_wait_event_info = &local_my_wait_event_info;
  * It is used to ensure that no duplicated entries are registered.
  *
  * The size of the hash table is based on the assumption that
- * WAIT_EVENT_EXTENSION_BASH_INIT_SIZE is enough for most cases, and it seems
+ * WAIT_EVENT_EXTENSION_HASH_INIT_SIZE is enough for most cases, and it seems
  * unlikely that the number of entries will reach
- * WAIT_EVENT_EXTENSION_BASH_MAX_SIZE.
+ * WAIT_EVENT_EXTENSION_HASH_MAX_SIZE.
  */
 static HTAB *WaitEventExtensionHashById;	/* find names from IDs */
 static HTAB *WaitEventExtensionHashByName;	/* find IDs from names */

src/include/access/xloginsert.h

@@ -37,6 +37,7 @@
 									 * will be skipped) */
 #define REGBUF_KEEP_DATA	0x10	/* include data even if a full-page image
 									 * is taken */
+#define REGBUF_NO_CHANGE	0x20	/* intentionally register clean buffer */
 
 /* prototypes for public functions in xloginsert.c: */
 extern void XLogBeginInsert(void);

src/include/backup/backup_manifest.h

@@ -39,7 +39,7 @@ extern void InitializeBackupManifest(backup_manifest_info *manifest,
 									 backup_manifest_option want_manifest,
 									 pg_checksum_type manifest_checksum_type);
 extern void AddFileToBackupManifest(backup_manifest_info *manifest,
-									const char *spcoid,
+									Oid spcoid,
 									const char *pathname, size_t size,
 									pg_time_t mtime,
 									pg_checksum_context *checksum_ctx);

src/include/backup/basebackup.h

@@ -27,7 +27,7 @@
  */
 typedef struct
 {
-	char	   *oid;			/* tablespace's OID, as a decimal string */
+	Oid			oid;			/* tablespace's OID */
 	char	   *path;			/* full path to tablespace's directory */
 	char	   *rpath;			/* relative path if it's within PGDATA, else
 								 * NULL */

src/include/storage/bufmgr.h

@@ -179,6 +179,8 @@ extern Buffer ReadBufferWithoutRelcache(RelFileLocator rlocator,
 										bool permanent);
 extern void ReleaseBuffer(Buffer buffer);
 extern void UnlockReleaseBuffer(Buffer buffer);
+extern bool BufferIsExclusiveLocked(Buffer buffer);
+extern bool BufferIsDirty(Buffer buffer);
 extern void MarkBufferDirty(Buffer buffer);
 extern void IncrBufferRefCount(Buffer buffer);
 extern void CheckBufferIsPinnedOnce(Buffer buffer);

src/include/storage/reinit.h

@@ -20,8 +20,9 @@
 
 extern void ResetUnloggedRelations(int op);
 extern bool parse_filename_for_nontemp_relation(const char *name,
-												int *relnumchars,
+												RelFileNumber *relnumber,
-												ForkNumber *fork);
+												ForkNumber *fork,
+												unsigned *segno);
 
 #define UNLOGGED_RELATION_CLEANUP		0x0001
 #define UNLOGGED_RELATION_INIT			0x0002