Adjust WAL code so that checkpoints truncate the xlog at the previous

checkpoint's redo pointer, not its undo pointer, per discussion in
pghackers a few days ago.  No point in hanging onto undo information
until we have the ability to do something with it --- and this solves
a rather large problem with log space for long-running transactions.
Also, change all calls of write() to detect the case where write
returned a count less than requested, but failed to set errno.
Presume that this situation indicates ENOSPC, and give the appropriate
error message, rather than a random message associated with the previous
value of errno.

6 files changed, 76 insertions, 20 deletions

diff --git a/contrib/dbase/dbf.c b/contrib/dbase/dbf.c
index 94f0b165cac..9ed0c475d12 100644
--- a/contrib/dbase/dbf.c
+++ b/contrib/dbase/dbf.c
@@ -155,9 +155,8 @@ int	dbf_write_head(dbhead *dbh) {
 	put_short(head.dbh_hlen, dbh->db_hlen);
 	put_short(head.dbh_rlen, dbh->db_rlen);
 	
-	if (write(dbh->db_fd, &head, sizeof(dbf_header)) == -1 ) {
+	if (write(dbh->db_fd, &head, sizeof(dbf_header)) != sizeof(dbf_header))
 		return DBF_ERROR;
-	}
 
 	return 0;
 }
@@ -180,14 +179,12 @@ int dbf_put_fields(dbhead *dbh) {
 		field.dbf_flen = dbh->db_fields[t].db_flen;
 		field.dbf_dec = dbh->db_fields[t].db_dec;
 
-		if (write(dbh->db_fd, &field, sizeof(dbf_field)) == -1) {
+		if (write(dbh->db_fd, &field, sizeof(dbf_field)) != sizeof(dbf_field))
 			return DBF_ERROR;
-		}
 	}
 
-	if (write(dbh->db_fd, &end, 1) == -1) {
+	if (write(dbh->db_fd, &end, 1) != 1)
 		return DBF_ERROR;
-	}
 
 	return 0;
 }
@@ -457,15 +454,13 @@ int dbf_put_record(dbhead *dbh, field *rec, u_long where) {
 	  idx += rec[t].db_flen;
 	}
 
-	if (write(dbh->db_fd, data, dbh->db_rlen) == -1) {
+	if (write(dbh->db_fd, data, dbh->db_rlen) != dbh->db_rlen)
 		return DBF_ERROR;
-	}
 
 /* There's a 0x1A at the end of a dbf-file */
 	if (where == dbh->db_records) {
-		if (write(dbh->db_fd, &end, 1) == -1) {
+		if (write(dbh->db_fd, &end, 1) != 1)
 			return DBF_ERROR;
-		}
 	}
 
 	dbh->db_offset += dbh->db_rlen;
diff --git a/contrib/pg_resetxlog/pg_resetxlog.c b/contrib/pg_resetxlog/pg_resetxlog.c
index 90fd3136dd8..f8c81b5c55d 100644
--- a/contrib/pg_resetxlog/pg_resetxlog.c
+++ b/contrib/pg_resetxlog/pg_resetxlog.c
@@ -23,7 +23,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Header: /cvsroot/pgsql/contrib/pg_resetxlog/Attic/pg_resetxlog.c,v 1.4 2001/05/30 14:18:18 momjian Exp $
+ * $Header: /cvsroot/pgsql/contrib/pg_resetxlog/Attic/pg_resetxlog.c,v 1.5 2001/06/06 17:07:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -772,8 +772,12 @@ RewriteControlFile(void)
 		exit(1);
 	}
 
+	errno = 0;
 	if (write(fd, buffer, BLCKSZ) != BLCKSZ)
 	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		perror("RewriteControlFile failed to write pg_control file");
 		exit(1);
 	}
@@ -884,8 +888,12 @@ WriteEmptyXLOG(void)
 		exit(1);
 	}
 
+	errno = 0;
 	if (write(fd, buffer, BLCKSZ) != BLCKSZ)
 	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		perror("WriteEmptyXLOG: failed to write xlog file");
 		exit(1);
 	}
@@ -894,8 +902,11 @@ WriteEmptyXLOG(void)
 	memset(buffer, 0, BLCKSZ);
 	for (nbytes = BLCKSZ; nbytes < XLogSegSize; nbytes += BLCKSZ)
 	{
+		errno = 0;
 		if (write(fd, buffer, BLCKSZ) != BLCKSZ)
 		{
+			if (errno == 0)
+				errno = ENOSPC;
 			perror("WriteEmptyXLOG: failed to write xlog file");
 			exit(1);
 		}
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index abbdba9eb41..95e414c03c9 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Header: /cvsroot/pgsql/src/backend/access/transam/xlog.c,v 1.68 2001/06/03 14:53:56 petere Exp $
+ * $Header: /cvsroot/pgsql/src/backend/access/transam/xlog.c,v 1.69 2001/06/06 17:07:38 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1068,9 +1068,15 @@ XLogWrite(XLogwrtRqst WriteRqst)
 
 		/* OK to write the page */
 		from = XLogCtl->pages + Write->curridx * BLCKSZ;
+		errno = 0;
 		if (write(openLogFile, from, BLCKSZ) != BLCKSZ)
+		{
+			/* if write didn't set errno, assume problem is no disk space */
+			if (errno == 0)
+				errno = ENOSPC;
 			elog(STOP, "write of log file %u, segment %u, offset %u failed: %m",
 				 openLogId, openLogSeg, openLogOff);
+		}
 		openLogOff += BLCKSZ;
 
 		/*
@@ -1323,6 +1329,7 @@ XLogFileInit(uint32 log, uint32 seg,
 	MemSet(zbuffer, 0, sizeof(zbuffer));
 	for (nbytes = 0; nbytes < XLogSegSize; nbytes += sizeof(zbuffer))
 	{
+		errno = 0;
 		if ((int) write(fd, zbuffer, sizeof(zbuffer)) != (int) sizeof(zbuffer))
 		{
 			int			save_errno = errno;
@@ -1332,7 +1339,8 @@ XLogFileInit(uint32 log, uint32 seg,
 			 * space
 			 */
 			unlink(tmppath);
-			errno = save_errno;
+			/* if write didn't set errno, assume problem is no disk space */
+			errno = save_errno ? save_errno : ENOSPC;
 
 			elog(STOP, "ZeroFill failed to create or write %s: %m", tmppath);
 		}
@@ -1990,8 +1998,14 @@ WriteControlFile(void)
 		elog(STOP, "WriteControlFile: could not create control file (%s): %m",
 			 ControlFilePath);
 
+	errno = 0;
 	if (write(fd, buffer, BLCKSZ) != BLCKSZ)
+	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		elog(STOP, "WriteControlFile: write to control file failed: %m");
+	}
 
 	if (pg_fsync(fd) != 0)
 		elog(STOP, "WriteControlFile: fsync of control file failed: %m");
@@ -2109,8 +2123,14 @@ UpdateControlFile(void)
 	if (fd < 0)
 		elog(STOP, "could not open control file (%s): %m", ControlFilePath);
 
+	errno = 0;
 	if (write(fd, ControlFile, sizeof(ControlFileData)) != sizeof(ControlFileData))
+	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		elog(STOP, "write to control file failed: %m");
+	}
 
 	if (pg_fsync(fd) != 0)
 		elog(STOP, "fsync of control file failed: %m");
@@ -2248,8 +2268,14 @@ BootStrapXLOG(void)
 	use_existent = false;
 	openLogFile = XLogFileInit(0, 0, &use_existent, false);
 
+	errno = 0;
 	if (write(openLogFile, buffer, BLCKSZ) != BLCKSZ)
+	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		elog(STOP, "BootStrapXLOG failed to write log file: %m");
+	}
 
 	if (pg_fsync(openLogFile) != 0)
 		elog(STOP, "BootStrapXLOG failed to fsync log file: %m");
@@ -2852,15 +2878,22 @@ CreateCheckPoint(bool shutdown)
 		elog(STOP, "concurrent transaction log activity while database system is shutting down");
 
 	/*
-	 * Remember location of prior checkpoint's earliest info. Oldest item
-	 * is redo or undo, whichever is older; but watch out for case that
-	 * undo = 0.
+	 * Select point at which we can truncate the log, which we base on the
+	 * prior checkpoint's earliest info.
+	 *
+	 * With UNDO support: oldest item is redo or undo, whichever is older;
+	 * but watch out for case that undo = 0.
+	 *
+	 * Without UNDO support: just use the redo pointer.  This allows xlog
+	 * space to be freed much faster when there are long-running transactions.
 	 */
+#ifdef NOT_USED
 	if (ControlFile->checkPointCopy.undo.xrecoff != 0 &&
 		XLByteLT(ControlFile->checkPointCopy.undo,
 				 ControlFile->checkPointCopy.redo))
 		XLByteToSeg(ControlFile->checkPointCopy.undo, _logId, _logSeg);
 	else
+#endif
 		XLByteToSeg(ControlFile->checkPointCopy.redo, _logId, _logSeg);
 
 	/*
diff --git a/src/backend/storage/file/fd.c b/src/backend/storage/file/fd.c
index 06ee4bc2729..237ae0a4428 100644
--- a/src/backend/storage/file/fd.c
+++ b/src/backend/storage/file/fd.c
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/storage/file/fd.c,v 1.79 2001/05/30 14:15:26 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/storage/file/fd.c,v 1.80 2001/06/06 17:07:46 tgl Exp $
  *
  * NOTES:
  *
@@ -865,7 +865,14 @@ FileWrite(File file, char *buffer, int amount)
 			   VfdCache[file].seekPos, amount, buffer));
 
 	FileAccess(file);
+
+	errno = 0;
 	returnCode = write(VfdCache[file].fd, buffer, amount);
+
+	/* if write didn't set errno, assume problem is no disk space */
+	if (returnCode != amount && errno == 0)
+		errno = ENOSPC;
+
 	if (returnCode > 0)
 		VfdCache[file].seekPos += returnCode;
 	else
diff --git a/src/backend/storage/smgr/md.c b/src/backend/storage/smgr/md.c
index 36c0f6d43b4..762aceda73c 100644
--- a/src/backend/storage/smgr/md.c
+++ b/src/backend/storage/smgr/md.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/storage/smgr/md.c,v 1.84 2001/05/10 20:38:49 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/storage/smgr/md.c,v 1.85 2001/06/06 17:07:46 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -582,8 +582,12 @@ mdblindwrt(RelFileNode rnode,
 	status = SM_SUCCESS;
 
 	/* write and optionally sync the block */
+	errno = 0;
 	if (write(fd, buffer, BLCKSZ) != BLCKSZ)
 	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		elog(DEBUG, "mdblindwrt: write() failed: %m");
 		status = SM_FAIL;
 	}
diff --git a/src/backend/utils/init/miscinit.c b/src/backend/utils/init/miscinit.c
index 544b57ff1d6..91711b26c36 100644
--- a/src/backend/utils/init/miscinit.c
+++ b/src/backend/utils/init/miscinit.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/utils/init/miscinit.c,v 1.68 2001/06/01 20:27:41 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/utils/init/miscinit.c,v 1.69 2001/06/06 17:07:46 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -659,13 +659,15 @@ CreateLockFile(const char *filename, bool amPostmaster,
 	snprintf(buffer, sizeof(buffer), "%d\n%s\n",
 			 amPostmaster ? (int) my_pid : -((int) my_pid),
 			 DataDir);
+	errno = 0;
 	if (write(fd, buffer, strlen(buffer)) != strlen(buffer))
 	{
 		int			save_errno = errno;
 
 		close(fd);
 		unlink(filename);
-		errno = save_errno;
+		/* if write didn't set errno, assume problem is no disk space */
+		errno = save_errno ? save_errno : ENOSPC;
 		elog(FATAL, "Can't write lock file %s: %m", filename);
 	}
 	close(fd);
@@ -794,9 +796,13 @@ RecordSharedMemoryInLockFile(IpcMemoryKey shmKey, IpcMemoryId shmId)
 	 * update should appear atomic to onlookers.
 	 */
 	len = strlen(buffer);
+	errno = 0;
 	if (lseek(fd, (off_t) 0, SEEK_SET) != 0 ||
 		(int) write(fd, buffer, len) != len)
 	{
+		/* if write didn't set errno, assume problem is no disk space */
+		if (errno == 0)
+			errno = ENOSPC;
 		elog(DEBUG, "Failed to write %s: %m", directoryLockFile);
 		close(fd);
 		return;