Make WAL segment size configurable at initdb time.

For performance reasons a larger segment size than the default 16MB
can be useful. A larger segment size has two main benefits: Firstly,
in setups using archiving, it makes it easier to write scripts that
can keep up with higher amounts of WAL, secondly, the WAL has to be
written and synced to disk less frequently.

But at the same time large segment size are disadvantageous for
smaller databases. So far the segment size had to be configured at
compile time, often making it unrealistic to choose one fitting to a
particularly load. Therefore change it to a initdb time setting.

This includes a breaking changes to the xlogreader.h API, which now
requires the current segment size to be configured.  For that and
similar reasons a number of binaries had to be taught how to recognize
the current segment size.

Author: Beena Emerson, editorialized by Andres Freund
Reviewed-By: Andres Freund, David Steele, Kuntal Ghosh, Michael
    Paquier, Peter Eisentraut, Robert Hass, Tushar Ahuja
Discussion: https://postgr.es/m/CAOG9ApEAcQ--1ieKbhFzXSQPw_YLmepaa4hNdnY5+ZULpt81Mw@mail.gmail.com

configure

@@ -821,7 +821,6 @@ enable_tap_tests
 with_blocksize
 with_segsize
 with_wal_blocksize
-with_wal_segsize
 with_CC
 enable_depend
 enable_cassert
@@ -1518,8 +1517,6 @@ Optional Packages:
   --with-segsize=SEGSIZE  set table segment size in GB [1]
   --with-wal-blocksize=BLOCKSIZE
                           set WAL block size in kB [8]
-  --with-wal-segsize=SEGSIZE
-                          set WAL segment size in MB [16]
   --with-CC=CMD           set compiler (deprecated)
   --with-icu              build with ICU support
   --with-tcl              build Tcl modules (PL/Tcl)
@@ -3733,57 +3730,6 @@ cat >>confdefs.h <<_ACEOF
 _ACEOF
 
 
-#
-# WAL segment size
-#
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for WAL segment size" >&5
-$as_echo_n "checking for WAL segment size... " >&6; }
-
-
-
-# Check whether --with-wal-segsize was given.
-if test "${with_wal_segsize+set}" = set; then :
-  withval=$with_wal_segsize;
-  case $withval in
-    yes)
-      as_fn_error $? "argument required for --with-wal-segsize option" "$LINENO" 5
-      ;;
-    no)
-      as_fn_error $? "argument required for --with-wal-segsize option" "$LINENO" 5
-      ;;
-    *)
-      wal_segsize=$withval
-      ;;
-  esac
-
-else
-  wal_segsize=16
-fi
-
-
-case ${wal_segsize} in
-  1) ;;
-  2) ;;
-  4) ;;
-  8) ;;
- 16) ;;
- 32) ;;
- 64) ;;
- 128) ;;
- 256) ;;
- 512) ;;
- 1024) ;;
-  *) as_fn_error $? "Invalid WAL segment size. Allowed values are 1,2,4,8,16,32,64,128,256,512,1024." "$LINENO" 5
-esac
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: ${wal_segsize}MB" >&5
-$as_echo "${wal_segsize}MB" >&6; }
-
-
-cat >>confdefs.h <<_ACEOF
-#define XLOG_SEG_SIZE (${wal_segsize} * 1024 * 1024)
-_ACEOF
-
-
 #
 # C compiler
 #

configure.in

@@ -343,37 +343,6 @@ AC_DEFINE_UNQUOTED([XLOG_BLCKSZ], ${XLOG_BLCKSZ}, [
  Changing XLOG_BLCKSZ requires an initdb.
 ])
 
-#
-# WAL segment size
-#
-AC_MSG_CHECKING([for WAL segment size])
-PGAC_ARG_REQ(with, wal-segsize, [SEGSIZE], [set WAL segment size in MB [16]],
-             [wal_segsize=$withval],
-             [wal_segsize=16])
-case ${wal_segsize} in
-  1) ;;
-  2) ;;
-  4) ;;
-  8) ;;
- 16) ;;
- 32) ;;
- 64) ;;
- 128) ;;
- 256) ;;
- 512) ;;
- 1024) ;;
-  *) AC_MSG_ERROR([Invalid WAL segment size. Allowed values are 1,2,4,8,16,32,64,128,256,512,1024.])
-esac
-AC_MSG_RESULT([${wal_segsize}MB])
-
-AC_DEFINE_UNQUOTED([XLOG_SEG_SIZE], [(${wal_segsize} * 1024 * 1024)], [
- XLOG_SEG_SIZE is the size of a single WAL file.  This must be a power of 2
- and larger than XLOG_BLCKSZ (preferably, a great deal larger than
- XLOG_BLCKSZ).
-
- Changing XLOG_SEG_SIZE requires an initdb.
-])
-
 #
 # C compiler
 #

contrib/pg_standby/pg_standby.c

@@ -36,6 +36,8 @@
 
 const char *progname;
 
+int			WalSegSz = -1;
+
 /* Options and defaults */
 int			sleeptime = 5;		/* amount of time to sleep between file checks */
 int			waittime = -1;		/* how long we have been waiting, -1 no wait
@@ -100,6 +102,10 @@ int			nextWALFileType;
 
 struct stat stat_buf;
 
+static bool SetWALFileNameForCleanup(void);
+static bool SetWALSegSize(void);
+
+
 /* =====================================================================
  *
  *		  Customizable section
@@ -175,6 +181,35 @@ CustomizableNextWALFileReady(void)
 {
 	if (stat(WALFilePath, &stat_buf) == 0)
 	{
+		/*
+		 * If we've not seen any WAL segments, we don't know the WAL segment
+		 * size, which we need. If it looks like a WAL segment, determine size
+		 * of segments for the cluster.
+		 */
+		if (WalSegSz == -1 && IsXLogFileName(nextWALFileName))
+		{
+			if (SetWALSegSize())
+			{
+				/*
+				 * Successfully determined WAL segment size. Can compute
+				 * cleanup cutoff now.
+				 */
+				need_cleanup = SetWALFileNameForCleanup();
+				if (debug)
+				{
+					fprintf(stderr,
+							_("WAL segment size:     %d \n"), WalSegSz);
+					fprintf(stderr, "Keep archive history: ");
+
+					if (need_cleanup)
+						fprintf(stderr, "%s and later\n",
+								exclusiveCleanupFileName);
+					else
+						fprintf(stderr, "no cleanup required\n");
+				}
+			}
+		}
+
 		/*
 		 * If it's a backup file, return immediately. If it's a regular file
 		 * return only if it's the right size already.
@@ -184,7 +219,7 @@ CustomizableNextWALFileReady(void)
 			nextWALFileType = XLOG_BACKUP_LABEL;
 			return true;
 		}
-		else if (stat_buf.st_size == XLOG_SEG_SIZE)
+		else if (WalSegSz > 0 && stat_buf.st_size == WalSegSz)
 		{
 #ifdef WIN32
 
@@ -204,7 +239,7 @@ CustomizableNextWALFileReady(void)
 		/*
 		 * If still too small, wait until it is the correct size
 		 */
-		if (stat_buf.st_size > XLOG_SEG_SIZE)
+		if (WalSegSz > 0 && stat_buf.st_size > WalSegSz)
 		{
 			if (debug)
 			{
@@ -218,8 +253,6 @@ CustomizableNextWALFileReady(void)
 	return false;
 }
 
-#define MaxSegmentsPerLogFile ( 0xFFFFFFFF / XLOG_SEG_SIZE )
-
 static void
 CustomizableCleanupPriorWALFiles(void)
 {
@@ -315,6 +348,7 @@ SetWALFileNameForCleanup(void)
 	uint32		log_diff = 0,
 				seg_diff = 0;
 	bool		cleanup = false;
+	int			max_segments_per_logfile = (0xFFFFFFFF / WalSegSz);
 
 	if (restartWALFileName)
 	{
@@ -336,12 +370,12 @@ SetWALFileNameForCleanup(void)
 		sscanf(nextWALFileName, "%08X%08X%08X", &tli, &log, &seg);
 		if (tli > 0 && seg > 0)
 		{
-			log_diff = keepfiles / MaxSegmentsPerLogFile;
-			seg_diff = keepfiles % MaxSegmentsPerLogFile;
+			log_diff = keepfiles / max_segments_per_logfile;
+			seg_diff = keepfiles % max_segments_per_logfile;
 			if (seg_diff > seg)
 			{
 				log_diff++;
-				seg = MaxSegmentsPerLogFile - (seg_diff - seg);
+				seg = max_segments_per_logfile - (seg_diff - seg);
 			}
 			else
 				seg -= seg_diff;
@@ -364,6 +398,66 @@ SetWALFileNameForCleanup(void)
 	return cleanup;
 }
 
+/*
+ * Try to set the wal segment size from the WAL file specified by WALFilePath.
+ *
+ * Return true if size could be determined, false otherwise.
+ */
+static bool
+SetWALSegSize(void)
+{
+	bool		ret_val = false;
+	int			fd;
+	char	   *buf = (char *) malloc(XLOG_BLCKSZ);
+
+	Assert(WalSegSz == -1);
+
+	if ((fd = open(WALFilePath, O_RDWR, 0)) < 0)
+	{
+		fprintf(stderr, "%s: couldn't open WAL file \"%s\"\n",
+				progname, WALFilePath);
+		return false;
+	}
+	if (read(fd, buf, XLOG_BLCKSZ) == XLOG_BLCKSZ)
+	{
+		XLogLongPageHeader longhdr = (XLogLongPageHeader) buf;
+
+		WalSegSz = longhdr->xlp_seg_size;
+
+		if (IsValidWalSegSize(WalSegSz))
+		{
+			/* successfully retrieved WAL segment size */
+			ret_val = true;
+		}
+		else
+			fprintf(stderr,
+					"%s: WAL segment size must be a power of two between 1MB and 1GB, but the WAL file header specifies %d bytes\n",
+					progname, WalSegSz);
+		close(fd);
+	}
+	else
+	{
+		/*
+		 * Don't complain loudly, this is to be expected for segments being
+		 * created.
+		 */
+		if (errno != 0)
+		{
+			if (debug)
+				fprintf(stderr, "could not read file \"%s\": %s",
+						WALFilePath, strerror(errno));
+		}
+		else
+		{
+			if (debug)
+				fprintf(stderr, "not enough data in file \"%s\"", WALFilePath);
+		}
+	}
+
+	fflush(stderr);
+	return ret_val;
+}
+
 /*
  * CheckForExternalTrigger()
  *
@@ -708,8 +802,6 @@ main(int argc, char **argv)
 
 	CustomizableInitialize();
 
-	need_cleanup = SetWALFileNameForCleanup();
-
 	if (debug)
 	{
 		fprintf(stderr, "Trigger file:         %s\n", triggerPath ? triggerPath : "<not set>");
@@ -721,11 +813,6 @@ main(int argc, char **argv)
 		fprintf(stderr, "Max wait interval:    %d %s\n",
 				maxwaittime, (maxwaittime > 0 ? "seconds" : "forever"));
 		fprintf(stderr, "Command for restore:  %s\n", restoreCommand);
-		fprintf(stderr, "Keep archive history: ");
-		if (need_cleanup)
-			fprintf(stderr, "%s and later\n", exclusiveCleanupFileName);
-		else
-			fprintf(stderr, "no cleanup required\n");
 		fflush(stderr);
 	}
 

doc/src/sgml/backup.sgml

@@ -562,7 +562,7 @@ tar -cf backup.tar /usr/local/pgsql/data
     produces an indefinitely long sequence of WAL records.  The system
     physically divides this sequence into WAL <firstterm>segment
     files</>, which are normally 16MB apiece (although the segment size
-    can be altered when building <productname>PostgreSQL</>).  The segment
+    can be altered during <application>initdb</>).  The segment
     files are given numeric names that reflect their position in the
     abstract WAL sequence.  When not using WAL archiving, the system
     normally creates just a few segment files and then

doc/src/sgml/installation.sgml

@@ -1058,20 +1058,6 @@ su - postgres
        </listitem>
       </varlistentry>
 
-      <varlistentry>
-       <term><option>--with-wal-segsize=<replaceable>SEGSIZE</replaceable></option></term>
-       <listitem>
-        <para>
-         Set the <firstterm>WAL segment size</>, in megabytes.  This is
-         the size of each individual file in the WAL log.  It may be useful
-         to adjust this size to control the granularity of WAL log shipping.
-         The default size is 16 megabytes.
-         The value must be a power of 2 between 1 and 1024 (megabytes).
-         Note that changing this value requires an initdb.
-        </para>
-       </listitem>
-      </varlistentry>
-
       <varlistentry>
        <term><option>--with-wal-blocksize=<replaceable>BLOCKSIZE</replaceable></option></term>
        <listitem>

doc/src/sgml/ref/initdb.sgml

@@ -316,6 +316,21 @@ PostgreSQL documentation
      </varlistentry>
 
      <varlistentry>
+       <term><option>--wal-segsize=<replaceable>SEGSIZE</replaceable></option></term>
+       <listitem>
+        <para>
+         Set the <firstterm>WAL segment size</>, in megabytes.  This is
+         the size of each individual file in the WAL log.  It may be useful
+         to adjust this size to control the granularity of WAL log shipping.
+         This option can only be set during initialization, and cannot be
+         changed later.
+         The default size is 16 megabytes.
+         The value must be a power of 2 between 1 and 1024 (megabytes).
+        </para>
+       </listitem>
+      </varlistentry>
+
+      <varlistentry>
       <term><option>-X <replaceable class="parameter">directory</replaceable></option></term>
       <term><option>--waldir=<replaceable class="parameter">directory</replaceable></option></term>
       <listitem>

doc/src/sgml/wal.sgml

@@ -752,13 +752,12 @@
    <acronym>WAL</acronym> logs are stored in the directory
    <filename>pg_wal</filename> under the data directory, as a set of
    segment files, normally each 16 MB in size (but the size can be changed
-   by altering the <option>--with-wal-segsize</> configure option when
-   building the server).  Each segment is divided into pages, normally
-   8 kB each (this size can be changed via the <option>--with-wal-blocksize</>
-   configure option).  The log record headers are described in
-   <filename>access/xlogrecord.h</filename>; the record content is dependent
-   on the type of event that is being logged.  Segment files are given
-   ever-increasing numbers as names, starting at
+   by altering the <option>--wal-segsize</> initdb option).  Each segment is
+   divided into pages, normally 8 kB each (this size can be changed via the
+   <option>--with-wal-blocksize</> configure option).  The log record headers
+   are described in <filename>access/xlogrecord.h</filename>; the record
+   content is dependent on the type of event that is being logged.  Segment
+   files are given ever-increasing numbers as names, starting at
    <filename>000000010000000000000000</filename>.  The numbers do not wrap,
    but it will take a very, very long time to exhaust the
    available stock of numbers.

src/backend/access/transam/twophase.c

@@ -1299,7 +1299,8 @@ XlogReadTwoPhaseData(XLogRecPtr lsn, char **buf, int *len)
 	XLogReaderState *xlogreader;
 	char	   *errormsg;
 
-	xlogreader = XLogReaderAllocate(&read_local_xlog_page, NULL);
+	xlogreader = XLogReaderAllocate(wal_segment_size, &read_local_xlog_page,
+									NULL);
 	if (!xlogreader)
 		ereport(ERROR,
 				(errcode(ERRCODE_OUT_OF_MEMORY),

src/backend/access/transam/xlogarchive.c

@@ -134,13 +134,14 @@ RestoreArchivedFile(char *path, const char *xlogfname,
 	if (cleanupEnabled)
 	{
 		GetOldestRestartPoint(&restartRedoPtr, &restartTli);
-		XLByteToSeg(restartRedoPtr, restartSegNo);
-		XLogFileName(lastRestartPointFname, restartTli, restartSegNo);
+		XLByteToSeg(restartRedoPtr, restartSegNo, wal_segment_size);
+		XLogFileName(lastRestartPointFname, restartTli, restartSegNo,
+					 wal_segment_size);
 		/* we shouldn't need anything earlier than last restart point */
 		Assert(strcmp(lastRestartPointFname, xlogfname) <= 0);
 	}
 	else
-		XLogFileName(lastRestartPointFname, 0, 0L);
+		XLogFileName(lastRestartPointFname, 0, 0L, wal_segment_size);
 
 	/*
 	 * construct the command to be executed
@@ -347,8 +348,9 @@ ExecuteRecoveryCommand(char *command, char *commandName, bool failOnSignal)
 	 * archive, though there is no requirement to do so.
 	 */
 	GetOldestRestartPoint(&restartRedoPtr, &restartTli);
-	XLByteToSeg(restartRedoPtr, restartSegNo);
-	XLogFileName(lastRestartPointFname, restartTli, restartSegNo);
+	XLByteToSeg(restartRedoPtr, restartSegNo, wal_segment_size);
+	XLogFileName(lastRestartPointFname, restartTli, restartSegNo,
+				 wal_segment_size);
 
 	/*
 	 * construct the command to be executed
@@ -547,7 +549,7 @@ XLogArchiveNotifySeg(XLogSegNo segno)
 {
 	char		xlog[MAXFNAMELEN];
 
-	XLogFileName(xlog, ThisTimeLineID, segno);
+	XLogFileName(xlog, ThisTimeLineID, segno, wal_segment_size);
 	XLogArchiveNotify(xlog);
 }
 

src/backend/access/transam/xlogfuncs.c

@@ -489,8 +489,8 @@ pg_walfile_name_offset(PG_FUNCTION_ARGS)
 	/*
 	 * xlogfilename
 	 */
-	XLByteToPrevSeg(locationpoint, xlogsegno);
-	XLogFileName(xlogfilename, ThisTimeLineID, xlogsegno);
+	XLByteToPrevSeg(locationpoint, xlogsegno, wal_segment_size);
+	XLogFileName(xlogfilename, ThisTimeLineID, xlogsegno, wal_segment_size);
 
 	values[0] = CStringGetTextDatum(xlogfilename);
 	isnull[0] = false;
@@ -498,7 +498,7 @@ pg_walfile_name_offset(PG_FUNCTION_ARGS)
 	/*
 	 * offset
 	 */
-	xrecoff = locationpoint % XLogSegSize;
+	xrecoff = XLogSegmentOffset(locationpoint, wal_segment_size);
 
 	values[1] = UInt32GetDatum(xrecoff);
 	isnull[1] = false;
@@ -530,8 +530,8 @@ pg_walfile_name(PG_FUNCTION_ARGS)
 				 errmsg("recovery is in progress"),
 				 errhint("pg_walfile_name() cannot be executed during recovery.")));
 
-	XLByteToPrevSeg(locationpoint, xlogsegno);
-	XLogFileName(xlogfilename, ThisTimeLineID, xlogsegno);
+	XLByteToPrevSeg(locationpoint, xlogsegno, wal_segment_size);
+	XLogFileName(xlogfilename, ThisTimeLineID, xlogsegno, wal_segment_size);
 
 	PG_RETURN_TEXT_P(cstring_to_text(xlogfilename));
 }

src/backend/access/transam/xlogreader.c

@@ -64,7 +64,8 @@ report_invalid_record(XLogReaderState *state, const char *fmt,...)
  * Returns NULL if the xlogreader couldn't be allocated.
  */
 XLogReaderState *
-XLogReaderAllocate(XLogPageReadCB pagereadfunc, void *private_data)
+XLogReaderAllocate(int wal_segment_size, XLogPageReadCB pagereadfunc,
+				   void *private_data)
 {
 	XLogReaderState *state;
 
@@ -91,6 +92,7 @@ XLogReaderAllocate(XLogPageReadCB pagereadfunc, void *private_data)
 		return NULL;
 	}
 
+	state->wal_segment_size = wal_segment_size;
 	state->read_page = pagereadfunc;
 	/* system_identifier initialized to zeroes above */
 	state->private_data = private_data;
@@ -466,8 +468,8 @@ XLogReadRecord(XLogReaderState *state, XLogRecPtr RecPtr, char **errormsg)
 		(record->xl_info & ~XLR_INFO_MASK) == XLOG_SWITCH)
 	{
 		/* Pretend it extends to end of segment */
-		state->EndRecPtr += XLogSegSize - 1;
-		state->EndRecPtr -= state->EndRecPtr % XLogSegSize;
+		state->EndRecPtr += state->wal_segment_size - 1;
+		state->EndRecPtr -= XLogSegmentOffset(state->EndRecPtr, state->wal_segment_size);
 	}
 
 	if (DecodeXLogRecord(state, record, errormsg))
@@ -509,8 +511,8 @@ ReadPageInternal(XLogReaderState *state, XLogRecPtr pageptr, int reqLen)
 
 	Assert((pageptr % XLOG_BLCKSZ) == 0);
 
-	XLByteToSeg(pageptr, targetSegNo);
-	targetPageOff = (pageptr % XLogSegSize);
+	XLByteToSeg(pageptr, targetSegNo, state->wal_segment_size);
+	targetPageOff = XLogSegmentOffset(pageptr, state->wal_segment_size);
 
 	/* check whether we have all the requested data already */
 	if (targetSegNo == state->readSegNo && targetPageOff == state->readOff &&
@@ -719,16 +721,16 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 
 	Assert((recptr % XLOG_BLCKSZ) == 0);
 
-	XLByteToSeg(recptr, segno);
-	offset = recptr % XLogSegSize;
+	XLByteToSeg(recptr, segno, state->wal_segment_size);
+	offset = XLogSegmentOffset(recptr, state->wal_segment_size);
 
-	XLogSegNoOffsetToRecPtr(segno, offset, recaddr);
+	XLogSegNoOffsetToRecPtr(segno, offset, recaddr, state->wal_segment_size);
 
 	if (hdr->xlp_magic != XLOG_PAGE_MAGIC)
 	{
 		char		fname[MAXFNAMELEN];
 
-		XLogFileName(fname, state->readPageTLI, segno);
+		XLogFileName(fname, state->readPageTLI, segno, state->wal_segment_size);
 
 		report_invalid_record(state,
 							  "invalid magic number %04X in log segment %s, offset %u",
@@ -742,7 +744,7 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 	{
 		char		fname[MAXFNAMELEN];
 
-		XLogFileName(fname, state->readPageTLI, segno);
+		XLogFileName(fname, state->readPageTLI, segno, state->wal_segment_size);
 
 		report_invalid_record(state,
 							  "invalid info bits %04X in log segment %s, offset %u",
@@ -775,10 +777,10 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 								  fhdrident_str, sysident_str);
 			return false;
 		}
-		else if (longhdr->xlp_seg_size != XLogSegSize)
+		else if (longhdr->xlp_seg_size != state->wal_segment_size)
 		{
 			report_invalid_record(state,
-								  "WAL file is from different database system: incorrect XLOG_SEG_SIZE in page header");
+								  "WAL file is from different database system: incorrect segment size in page header");
 			return false;
 		}
 		else if (longhdr->xlp_xlog_blcksz != XLOG_BLCKSZ)
@@ -792,7 +794,7 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 	{
 		char		fname[MAXFNAMELEN];
 
-		XLogFileName(fname, state->readPageTLI, segno);
+		XLogFileName(fname, state->readPageTLI, segno, state->wal_segment_size);
 
 		/* hmm, first page of file doesn't have a long header? */
 		report_invalid_record(state,
@@ -807,7 +809,7 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 	{
 		char		fname[MAXFNAMELEN];
 
-		XLogFileName(fname, state->readPageTLI, segno);
+		XLogFileName(fname, state->readPageTLI, segno, state->wal_segment_size);
 
 		report_invalid_record(state,
 							  "unexpected pageaddr %X/%X in log segment %s, offset %u",
@@ -832,7 +834,7 @@ ValidXLogPageHeader(XLogReaderState *state, XLogRecPtr recptr,
 		{
 			char		fname[MAXFNAMELEN];
 
-			XLogFileName(fname, state->readPageTLI, segno);
+			XLogFileName(fname, state->readPageTLI, segno, state->wal_segment_size);
 
 			report_invalid_record(state,
 								  "out-of-sequence timeline ID %u (after %u) in log segment %s, offset %u",

src/backend/access/transam/xlogutils.c

@@ -654,7 +654,8 @@ XLogTruncateRelation(RelFileNode rnode, ForkNumber forkNum,
  * frontend).  Probably these should be merged at some point.
  */
 static void
-XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
+XLogRead(char *buf, int segsize, TimeLineID tli, XLogRecPtr startptr,
+		 Size count)
 {
 	char	   *p;
 	XLogRecPtr	recptr;
@@ -666,6 +667,8 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 	static TimeLineID sendTLI = 0;
 	static uint32 sendOff = 0;
 
+	Assert(segsize == wal_segment_size);
+
 	p = buf;
 	recptr = startptr;
 	nbytes = count;
@@ -676,10 +679,10 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 		int			segbytes;
 		int			readbytes;
 
-		startoff = recptr % XLogSegSize;
+		startoff = XLogSegmentOffset(recptr, segsize);
 
 		/* Do we need to switch to a different xlog segment? */
-		if (sendFile < 0 || !XLByteInSeg(recptr, sendSegNo) ||
+		if (sendFile < 0 || !XLByteInSeg(recptr, sendSegNo, segsize) ||
 			sendTLI != tli)
 		{
 			char		path[MAXPGPATH];
@@ -687,9 +690,9 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 			if (sendFile >= 0)
 				close(sendFile);
 
-			XLByteToSeg(recptr, sendSegNo);
+			XLByteToSeg(recptr, sendSegNo, segsize);
 
-			XLogFilePath(path, tli, sendSegNo);
+			XLogFilePath(path, tli, sendSegNo, segsize);
 
 			sendFile = BasicOpenFile(path, O_RDONLY | PG_BINARY, 0);
 
@@ -717,7 +720,7 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 			{
 				char		path[MAXPGPATH];
 
-				XLogFilePath(path, tli, sendSegNo);
+				XLogFilePath(path, tli, sendSegNo, segsize);
 
 				ereport(ERROR,
 						(errcode_for_file_access(),
@@ -728,8 +731,8 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 		}
 
 		/* How many bytes are within this segment? */
-		if (nbytes > (XLogSegSize - startoff))
-			segbytes = XLogSegSize - startoff;
+		if (nbytes > (segsize - startoff))
+			segbytes = segsize - startoff;
 		else
 			segbytes = nbytes;
 
@@ -740,7 +743,7 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 		{
 			char		path[MAXPGPATH];
 
-			XLogFilePath(path, tli, sendSegNo);
+			XLogFilePath(path, tli, sendSegNo, segsize);
 
 			ereport(ERROR,
 					(errcode_for_file_access(),
@@ -798,7 +801,8 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 void
 XLogReadDetermineTimeline(XLogReaderState *state, XLogRecPtr wantPage, uint32 wantLength)
 {
-	const XLogRecPtr lastReadPage = state->readSegNo * XLogSegSize + state->readOff;
+	const XLogRecPtr lastReadPage = state->readSegNo *
+		state->wal_segment_size + state->readOff;
 
 	Assert(wantPage != InvalidXLogRecPtr && wantPage % XLOG_BLCKSZ == 0);
 	Assert(wantLength <= XLOG_BLCKSZ);
@@ -842,7 +846,8 @@ XLogReadDetermineTimeline(XLogReaderState *state, XLogRecPtr wantPage, uint32 wa
 	if (state->currTLIValidUntil != InvalidXLogRecPtr &&
 		state->currTLI != ThisTimeLineID &&
 		state->currTLI != 0 &&
-		(wantPage + wantLength) / XLogSegSize < state->currTLIValidUntil / XLogSegSize)
+		((wantPage + wantLength) / state->wal_segment_size) <
+		(state->currTLIValidUntil / state->wal_segment_size))
 		return;
 
 	/*
@@ -864,9 +869,11 @@ XLogReadDetermineTimeline(XLogReaderState *state, XLogRecPtr wantPage, uint32 wa
 		 */
 		List	   *timelineHistory = readTimeLineHistory(ThisTimeLineID);
 
-		XLogRecPtr	endOfSegment = (((wantPage / XLogSegSize) + 1) * XLogSegSize) - 1;
+		XLogRecPtr	endOfSegment = (((wantPage / state->wal_segment_size) + 1)
+									* state->wal_segment_size) - 1;
 
-		Assert(wantPage / XLogSegSize == endOfSegment / XLogSegSize);
+		Assert(wantPage / state->wal_segment_size ==
+			   endOfSegment / state->wal_segment_size);
 
 		/*
 		 * Find the timeline of the last LSN on the segment containing
@@ -1014,7 +1021,8 @@ read_local_xlog_page(XLogReaderState *state, XLogRecPtr targetPagePtr,
 	 * as 'count', read the whole page anyway. It's guaranteed to be
 	 * zero-padded up to the page boundary if it's incomplete.
 	 */
-	XLogRead(cur_page, *pageTLI, targetPagePtr, XLOG_BLCKSZ);
+	XLogRead(cur_page, state->wal_segment_size, *pageTLI, targetPagePtr,
+			 XLOG_BLCKSZ);
 
 	/* number of valid bytes in the buffer */
 	return count;

src/backend/bootstrap/bootstrap.c

@@ -19,6 +19,7 @@
 
 #include "access/htup_details.h"
 #include "access/xact.h"
+#include "access/xlog_internal.h"
 #include "bootstrap/bootstrap.h"
 #include "catalog/index.h"
 #include "catalog/pg_collation.h"
@@ -222,7 +223,7 @@ AuxiliaryProcessMain(int argc, char *argv[])
 	/* If no -x argument, we are a CheckerProcess */
 	MyAuxProcType = CheckerProcess;
 
-	while ((flag = getopt(argc, argv, "B:c:d:D:Fkr:x:-:")) != -1)
+	while ((flag = getopt(argc, argv, "B:c:d:D:Fkr:x:X:-:")) != -1)
 	{
 		switch (flag)
 		{
@@ -257,6 +258,18 @@ AuxiliaryProcessMain(int argc, char *argv[])
 			case 'x':
 				MyAuxProcType = atoi(optarg);
 				break;
+			case 'X':
+				{
+					int			WalSegSz = strtoul(optarg, NULL, 0);
+
+					if (!IsValidWalSegSize(WalSegSz))
+						ereport(ERROR,
+								(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+								 errmsg("-X requires a power of 2 value between 1MB and 1GB")));
+					SetConfigOption("wal_segment_size", optarg, PGC_INTERNAL,
+									PGC_S_OVERRIDE);
+				}
+				break;
 			case 'c':
 			case '-':
 				{

src/backend/postmaster/checkpointer.c

@@ -624,7 +624,7 @@ CheckArchiveTimeout(void)
 			 * If the returned pointer points exactly to a segment boundary,
 			 * assume nothing happened.
 			 */
-			if ((switchpoint % XLogSegSize) != 0)
+			if (XLogSegmentOffset(switchpoint, wal_segment_size) != 0)
 				elog(DEBUG1, "write-ahead log switch forced (archive_timeout=%d)",
 					 XLogArchiveTimeout);
 		}
@@ -782,7 +782,8 @@ IsCheckpointOnSchedule(double progress)
 		recptr = GetXLogReplayRecPtr(NULL);
 	else
 		recptr = GetInsertRecPtr();
-	elapsed_xlogs = (((double) (recptr - ckpt_start_recptr)) / XLogSegSize) / CheckPointSegments;
+	elapsed_xlogs = (((double) (recptr - ckpt_start_recptr)) /
+					 wal_segment_size) / CheckPointSegments;
 
 	if (progress < elapsed_xlogs)
 	{

src/backend/replication/basebackup.c

@@ -357,10 +357,10 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 		 * shouldn't be such files, but if there are, there's little harm in
 		 * including them.
 		 */
-		XLByteToSeg(startptr, startsegno);
-		XLogFileName(firstoff, ThisTimeLineID, startsegno);
-		XLByteToPrevSeg(endptr, endsegno);
-		XLogFileName(lastoff, ThisTimeLineID, endsegno);
+		XLByteToSeg(startptr, startsegno, wal_segment_size);
+		XLogFileName(firstoff, ThisTimeLineID, startsegno, wal_segment_size);
+		XLByteToPrevSeg(endptr, endsegno, wal_segment_size);
+		XLogFileName(lastoff, ThisTimeLineID, endsegno, wal_segment_size);
 
 		dir = AllocateDir("pg_wal");
 		if (!dir)
@@ -415,12 +415,13 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 		 * Sanity check: the first and last segment should cover startptr and
 		 * endptr, with no gaps in between.
 		 */
-		XLogFromFileName(walFiles[0], &tli, &segno);
+		XLogFromFileName(walFiles[0], &tli, &segno, wal_segment_size);
 		if (segno != startsegno)
 		{
 			char		startfname[MAXFNAMELEN];
 
-			XLogFileName(startfname, ThisTimeLineID, startsegno);
+			XLogFileName(startfname, ThisTimeLineID, startsegno,
+						 wal_segment_size);
 			ereport(ERROR,
 					(errmsg("could not find WAL file \"%s\"", startfname)));
 		}
@@ -429,12 +430,13 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 			XLogSegNo	currsegno = segno;
 			XLogSegNo	nextsegno = segno + 1;
 
-			XLogFromFileName(walFiles[i], &tli, &segno);
+			XLogFromFileName(walFiles[i], &tli, &segno, wal_segment_size);
 			if (!(nextsegno == segno || currsegno == segno))
 			{
 				char		nextfname[MAXFNAMELEN];
 
-				XLogFileName(nextfname, ThisTimeLineID, nextsegno);
+				XLogFileName(nextfname, ThisTimeLineID, nextsegno,
+							 wal_segment_size);
 				ereport(ERROR,
 						(errmsg("could not find WAL file \"%s\"", nextfname)));
 			}
@@ -443,7 +445,7 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 		{
 			char		endfname[MAXFNAMELEN];
 
-			XLogFileName(endfname, ThisTimeLineID, endsegno);
+			XLogFileName(endfname, ThisTimeLineID, endsegno, wal_segment_size);
 			ereport(ERROR,
 					(errmsg("could not find WAL file \"%s\"", endfname)));
 		}
@@ -457,7 +459,7 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 			pgoff_t		len = 0;
 
 			snprintf(pathbuf, MAXPGPATH, XLOGDIR "/%s", walFiles[i]);
-			XLogFromFileName(walFiles[i], &tli, &segno);
+			XLogFromFileName(walFiles[i], &tli, &segno, wal_segment_size);
 
 			fp = AllocateFile(pathbuf, "rb");
 			if (fp == NULL)
@@ -479,7 +481,7 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 						(errcode_for_file_access(),
 						 errmsg("could not stat file \"%s\": %m",
 								pathbuf)));
-			if (statbuf.st_size != XLogSegSize)
+			if (statbuf.st_size != wal_segment_size)
 			{
 				CheckXLogRemoved(segno, tli);
 				ereport(ERROR,
@@ -490,7 +492,9 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 			/* send the WAL file itself */
 			_tarWriteHeader(pathbuf, NULL, &statbuf, false);
 
-			while ((cnt = fread(buf, 1, Min(sizeof(buf), XLogSegSize - len), fp)) > 0)
+			while ((cnt = fread(buf, 1,
+								Min(sizeof(buf), wal_segment_size - len),
+								fp)) > 0)
 			{
 				CheckXLogRemoved(segno, tli);
 				/* Send the chunk as a CopyData message */
@@ -501,11 +505,11 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 				len += cnt;
 				throttle(cnt);
 
-				if (len == XLogSegSize)
+				if (len == wal_segment_size)
 					break;
 			}
 
-			if (len != XLogSegSize)
+			if (len != wal_segment_size)
 			{
 				CheckXLogRemoved(segno, tli);
 				ereport(ERROR,
@@ -513,7 +517,7 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 						 errmsg("unexpected WAL file size \"%s\"", walFiles[i])));
 			}
 
-			/* XLogSegSize is a multiple of 512, so no need for padding */
+			/* wal_segment_size is a multiple of 512, so no need for padding */
 
 			FreeFile(fp);
 

src/backend/replication/logical/logical.c

@@ -163,7 +163,7 @@ StartupDecodingContext(List *output_plugin_options,
 
 	ctx->slot = slot;
 
-	ctx->reader = XLogReaderAllocate(read_page, ctx);
+	ctx->reader = XLogReaderAllocate(wal_segment_size, read_page, ctx);
 	if (!ctx->reader)
 		ereport(ERROR,
 				(errcode(ERRCODE_OUT_OF_MEMORY),

src/backend/replication/logical/reorderbuffer.c

@@ -2083,15 +2083,16 @@ ReorderBufferSerializeTXN(ReorderBuffer *rb, ReorderBufferTXN *txn)
 		 * store in segment in which it belongs by start lsn, don't split over
 		 * multiple segments tho
 		 */
-		if (fd == -1 || !XLByteInSeg(change->lsn, curOpenSegNo))
+		if (fd == -1 ||
+			!XLByteInSeg(change->lsn, curOpenSegNo, wal_segment_size))
 		{
 			XLogRecPtr	recptr;
 
 			if (fd != -1)
 				CloseTransientFile(fd);
 
-			XLByteToSeg(change->lsn, curOpenSegNo);
-			XLogSegNoOffsetToRecPtr(curOpenSegNo, 0, recptr);
+			XLByteToSeg(change->lsn, curOpenSegNo, wal_segment_size);
+			XLogSegNoOffsetToRecPtr(curOpenSegNo, 0, recptr, wal_segment_size);
 
 			/*
 			 * No need to care about TLIs here, only used during a single run,
@@ -2319,7 +2320,7 @@ ReorderBufferRestoreChanges(ReorderBuffer *rb, ReorderBufferTXN *txn,
 	txn->nentries_mem = 0;
 	Assert(dlist_is_empty(&txn->changes));
 
-	XLByteToSeg(txn->final_lsn, last_segno);
+	XLByteToSeg(txn->final_lsn, last_segno, wal_segment_size);
 
 	while (restored < max_changes_in_memory && *segno <= last_segno)
 	{
@@ -2334,11 +2335,11 @@ ReorderBufferRestoreChanges(ReorderBuffer *rb, ReorderBufferTXN *txn,
 			/* first time in */
 			if (*segno == 0)
 			{
-				XLByteToSeg(txn->first_lsn, *segno);
+				XLByteToSeg(txn->first_lsn, *segno, wal_segment_size);
 			}
 
 			Assert(*segno != 0 || dlist_is_empty(&txn->changes));
-			XLogSegNoOffsetToRecPtr(*segno, 0, recptr);
+			XLogSegNoOffsetToRecPtr(*segno, 0, recptr, wal_segment_size);
 
 			/*
 			 * No need to care about TLIs here, only used during a single run,
@@ -2575,8 +2576,8 @@ ReorderBufferRestoreCleanup(ReorderBuffer *rb, ReorderBufferTXN *txn)
 	Assert(txn->first_lsn != InvalidXLogRecPtr);
 	Assert(txn->final_lsn != InvalidXLogRecPtr);
 
-	XLByteToSeg(txn->first_lsn, first);
-	XLByteToSeg(txn->final_lsn, last);
+	XLByteToSeg(txn->first_lsn, first, wal_segment_size);
+	XLByteToSeg(txn->final_lsn, last, wal_segment_size);
 
 	/* iterate over all possible filenames, and delete them */
 	for (cur = first; cur <= last; cur++)
@@ -2584,7 +2585,7 @@ ReorderBufferRestoreCleanup(ReorderBuffer *rb, ReorderBufferTXN *txn)
 		char		path[MAXPGPATH];
 		XLogRecPtr	recptr;
 
-		XLogSegNoOffsetToRecPtr(cur, 0, recptr);
+		XLogSegNoOffsetToRecPtr(cur, 0, recptr, wal_segment_size);
 
 		sprintf(path, "pg_replslot/%s/xid-%u-lsn-%X-%X.snap",
 				NameStr(MyReplicationSlot->data.name), txn->xid,

src/backend/replication/slot.c

@@ -1039,7 +1039,7 @@ ReplicationSlotReserveWal(void)
 		 * the new restart_lsn above, so normally we should never need to loop
 		 * more than twice.
 		 */
-		XLByteToSeg(slot->data.restart_lsn, segno);
+		XLByteToSeg(slot->data.restart_lsn, segno, wal_segment_size);
 		if (XLogGetLastRemovedSegno() < segno)
 			break;
 	}

src/backend/replication/walreceiver.c

@@ -613,7 +613,7 @@ WalReceiverMain(void)
 			 * Create .done file forcibly to prevent the streamed segment from
 			 * being archived later.
 			 */
-			XLogFileName(xlogfname, recvFileTLI, recvSegNo);
+			XLogFileName(xlogfname, recvFileTLI, recvSegNo, wal_segment_size);
 			if (XLogArchiveMode != ARCHIVE_MODE_ALWAYS)
 				XLogArchiveForceDone(xlogfname);
 			else
@@ -943,7 +943,7 @@ XLogWalRcvWrite(char *buf, Size nbytes, XLogRecPtr recptr)
 	{
 		int			segbytes;
 
-		if (recvFile < 0 || !XLByteInSeg(recptr, recvSegNo))
+		if (recvFile < 0 || !XLByteInSeg(recptr, recvSegNo, wal_segment_size))
 		{
 			bool		use_existent;
 
@@ -972,7 +972,7 @@ XLogWalRcvWrite(char *buf, Size nbytes, XLogRecPtr recptr)
 				 * Create .done file forcibly to prevent the streamed segment
 				 * from being archived later.
 				 */
-				XLogFileName(xlogfname, recvFileTLI, recvSegNo);
+				XLogFileName(xlogfname, recvFileTLI, recvSegNo, wal_segment_size);
 				if (XLogArchiveMode != ARCHIVE_MODE_ALWAYS)
 					XLogArchiveForceDone(xlogfname);
 				else
@@ -981,7 +981,7 @@ XLogWalRcvWrite(char *buf, Size nbytes, XLogRecPtr recptr)
 			recvFile = -1;
 
 			/* Create/use new log file */
-			XLByteToSeg(recptr, recvSegNo);
+			XLByteToSeg(recptr, recvSegNo, wal_segment_size);
 			use_existent = true;
 			recvFile = XLogFileInit(recvSegNo, &use_existent, true);
 			recvFileTLI = ThisTimeLineID;
@@ -989,10 +989,10 @@ XLogWalRcvWrite(char *buf, Size nbytes, XLogRecPtr recptr)
 		}
 
 		/* Calculate the start offset of the received logs */
-		startoff = recptr % XLogSegSize;
+		startoff = XLogSegmentOffset(recptr, wal_segment_size);
 
-		if (startoff + nbytes > XLogSegSize)
-			segbytes = XLogSegSize - startoff;
+		if (startoff + nbytes > wal_segment_size)
+			segbytes = wal_segment_size - startoff;
 		else
 			segbytes = nbytes;
 

src/backend/replication/walreceiverfuncs.c

@@ -233,8 +233,8 @@ RequestXLogStreaming(TimeLineID tli, XLogRecPtr recptr, const char *conninfo,
 	 * being created by XLOG streaming, which might cause trouble later on if
 	 * the segment is e.g archived.
 	 */
-	if (recptr % XLogSegSize != 0)
-		recptr -= recptr % XLogSegSize;
+	if (XLogSegmentOffset(recptr, wal_segment_size) != 0)
+		recptr -= XLogSegmentOffset(recptr, wal_segment_size);
 
 	SpinLockAcquire(&walrcv->mutex);
 

src/backend/replication/walsender.c

@@ -2316,9 +2316,9 @@ retry:
 		int			segbytes;
 		int			readbytes;
 
-		startoff = recptr % XLogSegSize;
+		startoff = XLogSegmentOffset(recptr, wal_segment_size);
 
-		if (sendFile < 0 || !XLByteInSeg(recptr, sendSegNo))
+		if (sendFile < 0 || !XLByteInSeg(recptr, sendSegNo, wal_segment_size))
 		{
 			char		path[MAXPGPATH];
 
@@ -2326,7 +2326,7 @@ retry:
 			if (sendFile >= 0)
 				close(sendFile);
 
-			XLByteToSeg(recptr, sendSegNo);
+			XLByteToSeg(recptr, sendSegNo, wal_segment_size);
 
 			/*-------
 			 * When reading from a historic timeline, and there is a timeline
@@ -2359,12 +2359,12 @@ retry:
 			{
 				XLogSegNo	endSegNo;
 
-				XLByteToSeg(sendTimeLineValidUpto, endSegNo);
+				XLByteToSeg(sendTimeLineValidUpto, endSegNo, wal_segment_size);
 				if (sendSegNo == endSegNo)
 					curFileTimeLine = sendTimeLineNextTLI;
 			}
 
-			XLogFilePath(path, curFileTimeLine, sendSegNo);
+			XLogFilePath(path, curFileTimeLine, sendSegNo, wal_segment_size);
 
 			sendFile = BasicOpenFile(path, O_RDONLY | PG_BINARY, 0);
 			if (sendFile < 0)
@@ -2401,8 +2401,8 @@ retry:
 		}
 
 		/* How many bytes are within this segment? */
-		if (nbytes > (XLogSegSize - startoff))
-			segbytes = XLogSegSize - startoff;
+		if (nbytes > (wal_segment_size - startoff))
+			segbytes = wal_segment_size - startoff;
 		else
 			segbytes = nbytes;
 
@@ -2433,7 +2433,7 @@ retry:
 	 * read() succeeds in that case, but the data we tried to read might
 	 * already have been overwritten with new WAL records.
 	 */
-	XLByteToSeg(startptr, segno);
+	XLByteToSeg(startptr, segno, wal_segment_size);
 	CheckXLogRemoved(segno, ThisTimeLineID);
 
 	/*

src/backend/utils/misc/guc.c

@@ -514,7 +514,6 @@ static int	block_size;
 static int	segment_size;
 static int	wal_block_size;
 static bool data_checksums;
-static int	wal_segment_size;
 static bool integer_datetimes;
 static bool assert_enabled;
 
@@ -714,9 +713,6 @@ typedef struct
 #if XLOG_BLCKSZ < 1024 || XLOG_BLCKSZ > (1024*1024)
 #error XLOG_BLCKSZ must be between 1KB and 1MB
 #endif
-#if XLOG_SEG_SIZE < (1024*1024) || XLOG_SEG_SIZE > (1024*1024*1024)
-#error XLOG_SEG_SIZE must be between 1MB and 1GB
-#endif
 
 static const char *memory_units_hint = gettext_noop("Valid units for this parameter are \"kB\", \"MB\", \"GB\", and \"TB\".");
 
@@ -2264,7 +2260,8 @@ static struct config_int ConfigureNamesInt[] =
 			GUC_UNIT_MB
 		},
 		&min_wal_size_mb,
-		5 * (XLOG_SEG_SIZE / (1024 * 1024)), 2, MAX_KILOBYTES,
+		DEFAULT_MIN_WAL_SEGS * (DEFAULT_XLOG_SEG_SIZE / (1024 * 1024)),
+		2, MAX_KILOBYTES,
 		NULL, NULL, NULL
 	},
 
@@ -2275,7 +2272,8 @@ static struct config_int ConfigureNamesInt[] =
 			GUC_UNIT_MB
 		},
 		&max_wal_size_mb,
-		64 * (XLOG_SEG_SIZE / (1024 * 1024)), 2, MAX_KILOBYTES,
+		DEFAULT_MAX_WAL_SEGS * (DEFAULT_XLOG_SEG_SIZE / (1024 * 1024)),
+		2, MAX_KILOBYTES,
 		NULL, assign_max_wal_size, NULL
 	},
 
@@ -2637,14 +2635,14 @@ static struct config_int ConfigureNamesInt[] =
 
 	{
 		{"wal_segment_size", PGC_INTERNAL, PRESET_OPTIONS,
-			gettext_noop("Shows the number of pages per write ahead log segment."),
+			gettext_noop("Shows the size of write ahead log segments."),
 			NULL,
-			GUC_UNIT_XBLOCKS | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE
+			GUC_UNIT_BYTE | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE
 		},
 		&wal_segment_size,
-		(XLOG_SEG_SIZE / XLOG_BLCKSZ),
-		(XLOG_SEG_SIZE / XLOG_BLCKSZ),
-		(XLOG_SEG_SIZE / XLOG_BLCKSZ),
+		DEFAULT_XLOG_SEG_SIZE,
+		WalSegMinSize,
+		WalSegMaxSize,
 		NULL, NULL, NULL
 	},
 

src/backend/utils/misc/pg_controldata.c

@@ -141,8 +141,9 @@ pg_control_checkpoint(PG_FUNCTION_ARGS)
 	 * Calculate name of the WAL file containing the latest checkpoint's REDO
 	 * start point.
 	 */
-	XLByteToSeg(ControlFile->checkPointCopy.redo, segno);
-	XLogFileName(xlogfilename, ControlFile->checkPointCopy.ThisTimeLineID, segno);
+	XLByteToSeg(ControlFile->checkPointCopy.redo, segno, wal_segment_size);
+	XLogFileName(xlogfilename, ControlFile->checkPointCopy.ThisTimeLineID,
+				 segno, wal_segment_size);
 
 	/* Populate the values and null arrays */
 	values[0] = LSNGetDatum(ControlFile->checkPoint);

src/backend/utils/misc/postgresql.conf.sample

@@ -234,7 +234,7 @@
 
 #max_wal_senders = 10		# max number of walsender processes
 				# (change requires restart)
-#wal_keep_segments = 0		# in logfile segments, 16MB each; 0 disables
+#wal_keep_segments = 0		# in logfile segments; 0 disables
 #wal_sender_timeout = 60s	# in milliseconds; 0 disables
 
 #max_replication_slots = 10	# max number of replication slots

src/bin/initdb/initdb.c

@@ -59,6 +59,7 @@
 #include "sys/mman.h"
 #endif
 
+#include "access/xlog_internal.h"
 #include "catalog/catalog.h"
 #include "catalog/pg_authid.h"
 #include "catalog/pg_class.h"
@@ -141,6 +142,8 @@ static bool sync_only = false;
 static bool show_setting = false;
 static bool data_checksums = false;
 static char *xlog_dir = NULL;
+static char *str_wal_segment_size_mb = NULL;
+static int	wal_segment_size_mb;
 
 
 /* internal vars */
@@ -999,6 +1002,23 @@ test_config_settings(void)
 	printf("%s\n", dynamic_shared_memory_type);
 }
 
+/*
+ * Calculate the default wal_size with a "pretty" unit.
+ */
+static char *
+pretty_wal_size(int segment_count)
+{
+	int			sz = wal_segment_size_mb * segment_count;
+	char	   *result = pg_malloc(10);
+
+	if ((sz % 1024) == 0)
+		snprintf(result, 10, "%dGB", sz / 1024);
+	else
+		snprintf(result, 10, "%dMB", sz);
+
+	return result;
+}
+
 /*
  * set up all the config files
  */
@@ -1043,6 +1063,15 @@ setup_config(void)
 	conflines = replace_token(conflines, "#port = 5432", repltok);
 #endif
 
+	/* set default max_wal_size and min_wal_size */
+	snprintf(repltok, sizeof(repltok), "min_wal_size = %s",
+			 pretty_wal_size(DEFAULT_MIN_WAL_SEGS));
+	conflines = replace_token(conflines, "#min_wal_size = 80MB", repltok);
+
+	snprintf(repltok, sizeof(repltok), "max_wal_size = %s",
+			 pretty_wal_size(DEFAULT_MAX_WAL_SEGS));
+	conflines = replace_token(conflines, "#max_wal_size = 1GB", repltok);
+
 	snprintf(repltok, sizeof(repltok), "lc_messages = '%s'",
 			 escape_quotes(lc_messages));
 	conflines = replace_token(conflines, "#lc_messages = 'C'", repltok);
@@ -1352,8 +1381,9 @@ bootstrap_template1(void)
 	unsetenv("PGCLIENTENCODING");
 
 	snprintf(cmd, sizeof(cmd),
-			 "\"%s\" --boot -x1 %s %s %s",
+			 "\"%s\" --boot -x1 -X %u %s %s %s",
 			 backend_exec,
+			 wal_segment_size_mb * (1024 * 1024),
 			 data_checksums ? "-k" : "",
 			 boot_options,
 			 debug ? "-d 5" : "");
@@ -2293,6 +2323,7 @@ usage(const char *progname)
 	printf(_("  -U, --username=NAME       database superuser name\n"));
 	printf(_("  -W, --pwprompt            prompt for a password for the new superuser\n"));
 	printf(_("  -X, --waldir=WALDIR       location for the write-ahead log directory\n"));
+	printf(_("      --wal-segsize=SIZE    size of wal segment size\n"));
 	printf(_("\nLess commonly used options:\n"));
 	printf(_("  -d, --debug               generate lots of debugging output\n"));
 	printf(_("  -k, --data-checksums      use data page checksums\n"));
@@ -2983,6 +3014,7 @@ main(int argc, char *argv[])
 		{"no-sync", no_argument, NULL, 'N'},
 		{"sync-only", no_argument, NULL, 'S'},
 		{"waldir", required_argument, NULL, 'X'},
+		{"wal-segsize", required_argument, NULL, 12},
 		{"data-checksums", no_argument, NULL, 'k'},
 		{NULL, 0, NULL, 0}
 	};
@@ -3116,6 +3148,9 @@ main(int argc, char *argv[])
 			case 'X':
 				xlog_dir = pg_strdup(optarg);
 				break;
+			case 12:
+				str_wal_segment_size_mb = pg_strdup(optarg);
+				break;
 			default:
 				/* getopt_long already emitted a complaint */
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
@@ -3178,6 +3213,27 @@ main(int argc, char *argv[])
 
 	check_need_password(authmethodlocal, authmethodhost);
 
+	/* set wal segment size */
+	if (str_wal_segment_size_mb == NULL)
+		wal_segment_size_mb = (DEFAULT_XLOG_SEG_SIZE) / (1024 * 1024);
+	else
+	{
+		char	   *endptr;
+
+		/* check that the argument is a number */
+		wal_segment_size_mb = strtol(str_wal_segment_size_mb, &endptr, 10);
+
+		/* verify that wal segment size is valid */
+		if (*endptr != '\0' ||
+			!IsValidWalSegSize(wal_segment_size_mb * 1024 * 1024))
+		{
+			fprintf(stderr,
+					_("%s: --wal-segsize must be a power of two between 1 and 1024\n"),
+					progname);
+			exit(1);
+		}
+	}
+
 	get_restricted_token(progname);
 
 	setup_pgdata();

src/bin/pg_basebackup/pg_basebackup.c

@@ -26,6 +26,7 @@
 #include <zlib.h>
 #endif
 
+#include "access/xlog_internal.h"
 #include "common/file_utils.h"
 #include "common/string.h"
 #include "fe_utils/string_utils.h"
@@ -555,7 +556,7 @@ StartLogStreamer(char *startpos, uint32 timeline, char *sysidentifier)
 	}
 	param->startptr = ((uint64) hi) << 32 | lo;
 	/* Round off to even segment position */
-	param->startptr -= param->startptr % XLOG_SEG_SIZE;
+	param->startptr -= XLogSegmentOffset(param->startptr, WalSegSz);
 
 #ifndef WIN32
 	/* Create our background pipe */
@@ -2397,6 +2398,10 @@ main(int argc, char **argv)
 		exit(1);
 	}
 
+	/* determine remote server's xlog segment size */
+	if (!RetrieveWalSegSize(conn))
+		disconnect_and_exit(1);
+
 	/* Create pg_wal symlink, if required */
 	if (xlog_dir)
 	{

src/bin/pg_basebackup/pg_receivewal.c

@@ -191,7 +191,7 @@ close_destination_dir(DIR *dest_dir, char *dest_folder)
 /*
  * Determine starting location for streaming, based on any existing xlog
  * segments in the directory. We start at the end of the last one that is
- * complete (size matches XLogSegSize), on the timeline with highest ID.
+ * complete (size matches wal segment size), on the timeline with highest ID.
  *
  * If there are no WAL files in the directory, returns InvalidXLogRecPtr.
  */
@@ -242,7 +242,7 @@ FindStreamingStart(uint32 *tli)
 		/*
 		 * Looks like an xlog file. Parse its position.
 		 */
-		XLogFromFileName(dirent->d_name, &tli, &segno);
+		XLogFromFileName(dirent->d_name, &tli, &segno, WalSegSz);
 
 		/*
 		 * Check that the segment has the right size, if it's supposed to be
@@ -267,7 +267,7 @@ FindStreamingStart(uint32 *tli)
 				disconnect_and_exit(1);
 			}
 
-			if (statbuf.st_size != XLOG_SEG_SIZE)
+			if (statbuf.st_size != WalSegSz)
 			{
 				fprintf(stderr,
 						_("%s: segment file \"%s\" has incorrect size %d, skipping\n"),
@@ -308,7 +308,7 @@ FindStreamingStart(uint32 *tli)
 			bytes_out = (buf[3] << 24) | (buf[2] << 16) |
 				(buf[1] << 8) | buf[0];
 
-			if (bytes_out != XLOG_SEG_SIZE)
+			if (bytes_out != WalSegSz)
 			{
 				fprintf(stderr,
 						_("%s: compressed segment file \"%s\" has incorrect uncompressed size %d, skipping\n"),
@@ -349,7 +349,7 @@ FindStreamingStart(uint32 *tli)
 		if (!high_ispartial)
 			high_segno++;
 
-		XLogSegNoOffsetToRecPtr(high_segno, 0, high_ptr);
+		XLogSegNoOffsetToRecPtr(high_segno, 0, high_ptr, WalSegSz);
 
 		*tli = high_tli;
 		return high_ptr;
@@ -410,7 +410,7 @@ StreamLog(void)
 	/*
 	 * Always start streaming at the beginning of a segment
 	 */
-	stream.startpos -= stream.startpos % XLOG_SEG_SIZE;
+	stream.startpos -= XLogSegmentOffset(stream.startpos, WalSegSz);
 
 	/*
 	 * Start the replication
@@ -689,6 +689,10 @@ main(int argc, char **argv)
 	if (!RunIdentifySystem(conn, NULL, NULL, NULL, &db_name))
 		disconnect_and_exit(1);
 
+	/* determine remote server's xlog segment size */
+	if (!RetrieveWalSegSize(conn))
+		disconnect_and_exit(1);
+
 	/*
 	 * Check that there is a database associated with connection, none should
 	 * be defined in this context.

src/bin/pg_basebackup/receivelog.c

@@ -95,17 +95,17 @@ open_walfile(StreamCtl *stream, XLogRecPtr startpoint)
 	ssize_t		size;
 	XLogSegNo	segno;
 
-	XLByteToSeg(startpoint, segno);
-	XLogFileName(current_walfile_name, stream->timeline, segno);
+	XLByteToSeg(startpoint, segno, WalSegSz);
+	XLogFileName(current_walfile_name, stream->timeline, segno, WalSegSz);
 
 	snprintf(fn, sizeof(fn), "%s%s", current_walfile_name,
 			 stream->partial_suffix ? stream->partial_suffix : "");
 
 	/*
 	 * When streaming to files, if an existing file exists we verify that it's
-	 * either empty (just created), or a complete XLogSegSize segment (in
-	 * which case it has been created and padded). Anything else indicates a
-	 * corrupt file.
+	 * either empty (just created), or a complete WalSegSz segment (in which
+	 * case it has been created and padded). Anything else indicates a corrupt
+	 * file.
 	 *
 	 * When streaming to tar, no file with this name will exist before, so we
 	 * never have to verify a size.
@@ -120,7 +120,7 @@ open_walfile(StreamCtl *stream, XLogRecPtr startpoint)
 					progname, fn, stream->walmethod->getlasterror());
 			return false;
 		}
-		if (size == XLogSegSize)
+		if (size == WalSegSz)
 		{
 			/* Already padded file. Open it for use */
 			f = stream->walmethod->open_for_write(current_walfile_name, stream->partial_suffix, 0);
@@ -154,7 +154,7 @@ open_walfile(StreamCtl *stream, XLogRecPtr startpoint)
 					ngettext("%s: write-ahead log file \"%s\" has %d byte, should be 0 or %d\n",
 							 "%s: write-ahead log file \"%s\" has %d bytes, should be 0 or %d\n",
 							 size),
-					progname, fn, (int) size, XLogSegSize);
+					progname, fn, (int) size, WalSegSz);
 			return false;
 		}
 		/* File existed and was empty, so fall through and open */
@@ -162,7 +162,8 @@ open_walfile(StreamCtl *stream, XLogRecPtr startpoint)
 
 	/* No file existed, so create one */
 
-	f = stream->walmethod->open_for_write(current_walfile_name, stream->partial_suffix, XLogSegSize);
+	f = stream->walmethod->open_for_write(current_walfile_name,
+										  stream->partial_suffix, WalSegSz);
 	if (f == NULL)
 	{
 		fprintf(stderr,
@@ -203,7 +204,7 @@ close_walfile(StreamCtl *stream, XLogRecPtr pos)
 
 	if (stream->partial_suffix)
 	{
-		if (currpos == XLOG_SEG_SIZE)
+		if (currpos == WalSegSz)
 			r = stream->walmethod->close(walfile, CLOSE_NORMAL);
 		else
 		{
@@ -231,7 +232,7 @@ close_walfile(StreamCtl *stream, XLogRecPtr pos)
 	 * new node. This is in line with walreceiver.c always doing a
 	 * XLogArchiveForceDone() after a complete segment.
 	 */
-	if (currpos == XLOG_SEG_SIZE && stream->mark_done)
+	if (currpos == WalSegSz && stream->mark_done)
 	{
 		/* writes error message if failed */
 		if (!mark_file_as_archived(stream, current_walfile_name))
@@ -676,7 +677,8 @@ ReceiveXlogStream(PGconn *conn, StreamCtl *stream)
 			 * start streaming at the beginning of a segment.
 			 */
 			stream->timeline = newtimeline;
-			stream->startpos = stream->startpos - (stream->startpos % XLOG_SEG_SIZE);
+			stream->startpos = stream->startpos -
+				XLogSegmentOffset(stream->startpos, WalSegSz);
 			continue;
 		}
 		else if (PQresultStatus(res) == PGRES_COMMAND_OK)
@@ -1111,7 +1113,7 @@ ProcessXLogDataMsg(PGconn *conn, StreamCtl *stream, char *copybuf, int len,
 	*blockpos = fe_recvint64(&copybuf[1]);
 
 	/* Extract WAL location for this block */
-	xlogoff = *blockpos % XLOG_SEG_SIZE;
+	xlogoff = XLogSegmentOffset(*blockpos, WalSegSz);
 
 	/*
 	 * Verify that the initial location in the stream matches where we think
@@ -1148,11 +1150,11 @@ ProcessXLogDataMsg(PGconn *conn, StreamCtl *stream, char *copybuf, int len,
 		int			bytes_to_write;
 
 		/*
-		 * If crossing a WAL boundary, only write up until we reach
-		 * XLOG_SEG_SIZE.
+		 * If crossing a WAL boundary, only write up until we reach wal
+		 * segment size.
 		 */
-		if (xlogoff + bytes_left > XLOG_SEG_SIZE)
-			bytes_to_write = XLOG_SEG_SIZE - xlogoff;
+		if (xlogoff + bytes_left > WalSegSz)
+			bytes_to_write = WalSegSz - xlogoff;
 		else
 			bytes_to_write = bytes_left;
 
@@ -1182,7 +1184,7 @@ ProcessXLogDataMsg(PGconn *conn, StreamCtl *stream, char *copybuf, int len,
 		xlogoff += bytes_to_write;
 
 		/* Did we reach the end of a WAL segment? */
-		if (*blockpos % XLOG_SEG_SIZE == 0)
+		if (XLogSegmentOffset(*blockpos, WalSegSz) == 0)
 		{
 			if (!close_walfile(stream, *blockpos))
 				/* Error message written in close_walfile() */

src/bin/pg_basebackup/streamutil.c

@@ -25,12 +25,18 @@
 #include "receivelog.h"
 #include "streamutil.h"
 
+#include "access/xlog_internal.h"
 #include "pqexpbuffer.h"
 #include "common/fe_memutils.h"
 #include "datatype/timestamp.h"
 
 #define ERRCODE_DUPLICATE_OBJECT  "42710"
 
+uint32		WalSegSz;
+
+/* SHOW command for replication connection was introduced in version 10 */
+#define MINIMUM_VERSION_FOR_SHOW_CMD 100000
+
 const char *progname;
 char	   *connection_string = NULL;
 char	   *dbhost = NULL;
@@ -231,6 +237,76 @@ GetConnection(void)
 	return tmpconn;
 }
 
+/*
+ * From version 10, explicitly set wal segment size using SHOW wal_segment_size
+ * since ControlFile is not accessible here.
+ */
+bool
+RetrieveWalSegSize(PGconn *conn)
+{
+	PGresult   *res;
+	char		xlog_unit[3];
+	int			xlog_val,
+				multiplier = 1;
+
+	/* check connection existence */
+	Assert(conn != NULL);
+
+	/* for previous versions set the default xlog seg size */
+	if (PQserverVersion(conn) < MINIMUM_VERSION_FOR_SHOW_CMD)
+	{
+		WalSegSz = DEFAULT_XLOG_SEG_SIZE;
+		return true;
+	}
+
+	res = PQexec(conn, "SHOW wal_segment_size");
+	if (PQresultStatus(res) != PGRES_TUPLES_OK)
+	{
+		fprintf(stderr, _("%s: could not send replication command \"%s\": %s\n"),
+				progname, "SHOW wal_segment_size", PQerrorMessage(conn));
+
+		PQclear(res);
+		return false;
+	}
+	if (PQntuples(res) != 1 || PQnfields(res) < 1)
+	{
+		fprintf(stderr,
+				_("%s: could not fetch WAL segment size: got %d rows and %d fields, expected %d rows and %d or more fields\n"),
+				progname, PQntuples(res), PQnfields(res), 1, 1);
+
+		PQclear(res);
+		return false;
+	}
+
+	/* fetch xlog value and unit from the result */
+	if (sscanf(PQgetvalue(res, 0, 0), "%d%s", &xlog_val, xlog_unit) != 2)
+	{
+		fprintf(stderr, _("%s: WAL segment size could not be parsed\n"),
+				progname);
+		return false;
+	}
+
+	/* set the multiplier based on unit to convert xlog_val to bytes */
+	if (strcmp(xlog_unit, "MB") == 0)
+		multiplier = 1024 * 1024;
+	else if (strcmp(xlog_unit, "GB") == 0)
+		multiplier = 1024 * 1024 * 1024;
+
+	/* convert and set WalSegSz */
+	WalSegSz = xlog_val * multiplier;
+
+	if (!IsValidWalSegSize(WalSegSz))
+	{
+		fprintf(stderr,
+				_("%s: WAL segment size must be a power of two between 1MB and 1GB, but the remote server reported a value of %d bytes\n"),
+				progname, WalSegSz);
+		return false;
+	}
+
+	PQclear(res);
+	return true;
+}
+
 /*
  * Run IDENTIFY_SYSTEM through a given connection and give back to caller
  * some result information if requested:

src/bin/pg_basebackup/streamutil.h

@@ -24,6 +24,7 @@ extern char *dbuser;
 extern char *dbport;
 extern char *dbname;
 extern int	dbgetpassword;
+extern uint32 WalSegSz;
 
 /* Connection kept global so we can disconnect easily */
 extern PGconn *conn;
@@ -39,6 +40,7 @@ extern bool RunIdentifySystem(PGconn *conn, char **sysid,
 				  TimeLineID *starttli,
 				  XLogRecPtr *startpos,
 				  char **db_name);
+extern bool RetrieveWalSegSize(PGconn *conn);
 extern TimestampTz feGetCurrentTimestamp(void);
 extern void feTimestampDifference(TimestampTz start_time, TimestampTz stop_time,
 					  long *secs, int *microsecs);

src/bin/pg_controldata/pg_controldata.c

@@ -99,6 +99,7 @@ main(int argc, char *argv[])
 	char		xlogfilename[MAXFNAMELEN];
 	int			c;
 	int			i;
+	int			WalSegSz;
 
 	set_pglocale_pgservice(argv[0], PG_TEXTDOMAIN("pg_controldata"));
 
@@ -164,6 +165,15 @@ main(int argc, char *argv[])
 				 "Either the file is corrupt, or it has a different layout than this program\n"
 				 "is expecting.  The results below are untrustworthy.\n\n"));
 
+	/* set wal segment size */
+	WalSegSz = ControlFile->xlog_seg_size;
+
+	if (!IsValidWalSegSize(WalSegSz))
+		fprintf(stderr,
+				_("WARNING: WAL segment size specified, %d bytes, is not a power of two between 1MB and 1GB.\n"
+				  "The file is corrupt and the results below are untrustworthy.\n"),
+				WalSegSz);
+
 	/*
 	 * This slightly-chintzy coding will work as long as the control file
 	 * timestamps are within the range of time_t; that should be the case in
@@ -184,8 +194,9 @@ main(int argc, char *argv[])
 	 * Calculate name of the WAL file containing the latest checkpoint's REDO
 	 * start point.
 	 */
-	XLByteToSeg(ControlFile->checkPointCopy.redo, segno);
-	XLogFileName(xlogfilename, ControlFile->checkPointCopy.ThisTimeLineID, segno);
+	XLByteToSeg(ControlFile->checkPointCopy.redo, segno, WalSegSz);
+	XLogFileName(xlogfilename, ControlFile->checkPointCopy.ThisTimeLineID,
+				 segno, WalSegSz);
 
 	/*
 	 * Format system_identifier and mock_authentication_nonce separately to

src/bin/pg_resetwal/pg_resetwal.c

@@ -70,6 +70,7 @@ static MultiXactId set_mxid = 0;
 static MultiXactOffset set_mxoff = (MultiXactOffset) -1;
 static uint32 minXlogTli = 0;
 static XLogSegNo minXlogSegNo = 0;
+static int	WalSegSz;
 
 static void CheckDataVersion(void);
 static bool ReadControlFile(void);
@@ -94,6 +95,7 @@ main(int argc, char *argv[])
 	char	   *endptr;
 	char	   *endptr2;
 	char	   *DataDir = NULL;
+	char	   *log_fname = NULL;
 	int			fd;
 
 	set_pglocale_pgservice(argv[0], PG_TEXTDOMAIN("pg_resetwal"));
@@ -265,7 +267,12 @@ main(int argc, char *argv[])
 					fprintf(stderr, _("Try \"%s --help\" for more information.\n"), progname);
 					exit(1);
 				}
-				XLogFromFileName(optarg, &minXlogTli, &minXlogSegNo);
+
+				/*
+				 * XLogFromFileName requires wal segment size which is not yet
+				 * set. Hence wal details are set later on.
+				 */
+				log_fname = pg_strdup(optarg);
 				break;
 
 			default:
@@ -350,6 +357,9 @@ main(int argc, char *argv[])
 	if (!ReadControlFile())
 		GuessControlValues();
 
+	if (log_fname != NULL)
+		XLogFromFileName(log_fname, &minXlogTli, &minXlogSegNo, WalSegSz);
+
 	/*
 	 * Also look at existing segment files to set up newXlogSegNo
 	 */
@@ -573,18 +583,27 @@ ReadControlFile(void)
 					offsetof(ControlFileData, crc));
 		FIN_CRC32C(crc);
 
-		if (EQ_CRC32C(crc, ((ControlFileData *) buffer)->crc))
+		if (!EQ_CRC32C(crc, ((ControlFileData *) buffer)->crc))
 		{
-			/* Valid data... */
-			memcpy(&ControlFile, buffer, sizeof(ControlFile));
-			return true;
+			/* We will use the data but treat it as guessed. */
+			fprintf(stderr,
+					_("%s: pg_control exists but has invalid CRC; proceed with caution\n"),
+					progname);
+			guessed = true;
 		}
 
-		fprintf(stderr, _("%s: pg_control exists but has invalid CRC; proceed with caution\n"),
-				progname);
-		/* We will use the data anyway, but treat it as guessed. */
 		memcpy(&ControlFile, buffer, sizeof(ControlFile));
-		guessed = true;
+		WalSegSz = ControlFile.xlog_seg_size;
+
+		/* return false if WalSegSz is not valid */
+		if (!IsValidWalSegSize(WalSegSz))
+		{
+			fprintf(stderr,
+					_("%s: pg_control specifies invalid WAL segment size (%d bytes); proceed with caution \n"),
+					progname, WalSegSz);
+			guessed = true;
+		}
+
 		return true;
 	}
 
@@ -660,7 +679,7 @@ GuessControlValues(void)
 	ControlFile.blcksz = BLCKSZ;
 	ControlFile.relseg_size = RELSEG_SIZE;
 	ControlFile.xlog_blcksz = XLOG_BLCKSZ;
-	ControlFile.xlog_seg_size = XLOG_SEG_SIZE;
+	ControlFile.xlog_seg_size = DEFAULT_XLOG_SEG_SIZE;
 	ControlFile.nameDataLen = NAMEDATALEN;
 	ControlFile.indexMaxKeys = INDEX_MAX_KEYS;
 	ControlFile.toast_max_chunk_size = TOAST_MAX_CHUNK_SIZE;
@@ -773,7 +792,8 @@ PrintNewControlValues(void)
 	/* This will be always printed in order to keep format same. */
 	printf(_("\n\nValues to be changed:\n\n"));
 
-	XLogFileName(fname, ControlFile.checkPointCopy.ThisTimeLineID, newXlogSegNo);
+	XLogFileName(fname, ControlFile.checkPointCopy.ThisTimeLineID,
+				 newXlogSegNo, WalSegSz);
 	printf(_("First log segment after reset:        %s\n"), fname);
 
 	if (set_mxid != 0)
@@ -850,7 +870,7 @@ RewriteControlFile(void)
 	 * newXlogSegNo.
 	 */
 	XLogSegNoOffsetToRecPtr(newXlogSegNo, SizeOfXLogLongPHD,
-							ControlFile.checkPointCopy.redo);
+							ControlFile.checkPointCopy.redo, WalSegSz);
 	ControlFile.checkPointCopy.time = (pg_time_t) time(NULL);
 
 	ControlFile.state = DB_SHUTDOWNED;
@@ -877,7 +897,7 @@ RewriteControlFile(void)
 	ControlFile.max_locks_per_xact = 64;
 
 	/* Now we can force the recorded xlog seg size to the right thing. */
-	ControlFile.xlog_seg_size = XLogSegSize;
+	ControlFile.xlog_seg_size = WalSegSz;
 
 	/* Contents are protected with a CRC */
 	INIT_CRC32C(ControlFile.crc);
@@ -1014,7 +1034,7 @@ FindEndOfXLOG(void)
 	 * are in virgin territory.
 	 */
 	xlogbytepos = newXlogSegNo * ControlFile.xlog_seg_size;
-	newXlogSegNo = (xlogbytepos + XLogSegSize - 1) / XLogSegSize;
+	newXlogSegNo = (xlogbytepos + WalSegSz - 1) / WalSegSz;
 	newXlogSegNo++;
 }
 
@@ -1151,7 +1171,7 @@ WriteEmptyXLOG(void)
 	page->xlp_pageaddr = ControlFile.checkPointCopy.redo - SizeOfXLogLongPHD;
 	longpage = (XLogLongPageHeader) page;
 	longpage->xlp_sysid = ControlFile.system_identifier;
-	longpage->xlp_seg_size = XLogSegSize;
+	longpage->xlp_seg_size = WalSegSz;
 	longpage->xlp_xlog_blcksz = XLOG_BLCKSZ;
 
 	/* Insert the initial checkpoint record */
@@ -1176,7 +1196,8 @@ WriteEmptyXLOG(void)
 	record->xl_crc = crc;
 
 	/* Write the first page */
-	XLogFilePath(path, ControlFile.checkPointCopy.ThisTimeLineID, newXlogSegNo);
+	XLogFilePath(path, ControlFile.checkPointCopy.ThisTimeLineID,
+				 newXlogSegNo, WalSegSz);
 
 	unlink(path);
 
@@ -1202,7 +1223,7 @@ WriteEmptyXLOG(void)
 
 	/* Fill the rest of the file with zeroes */
 	memset(buffer, 0, XLOG_BLCKSZ);
-	for (nbytes = XLOG_BLCKSZ; nbytes < XLogSegSize; nbytes += XLOG_BLCKSZ)
+	for (nbytes = XLOG_BLCKSZ; nbytes < WalSegSz; nbytes += XLOG_BLCKSZ)
 	{
 		errno = 0;
 		if (write(fd, buffer, XLOG_BLCKSZ) != XLOG_BLCKSZ)

src/bin/pg_rewind/parsexlog.c

@@ -69,7 +69,8 @@ extractPageMap(const char *datadir, XLogRecPtr startpoint, int tliIndex,
 
 	private.datadir = datadir;
 	private.tliIndex = tliIndex;
-	xlogreader = XLogReaderAllocate(&SimpleXLogPageRead, &private);
+	xlogreader = XLogReaderAllocate(WalSegSz, &SimpleXLogPageRead,
+									&private);
 	if (xlogreader == NULL)
 		pg_fatal("out of memory\n");
 
@@ -122,7 +123,8 @@ readOneRecord(const char *datadir, XLogRecPtr ptr, int tliIndex)
 
 	private.datadir = datadir;
 	private.tliIndex = tliIndex;
-	xlogreader = XLogReaderAllocate(&SimpleXLogPageRead, &private);
+	xlogreader = XLogReaderAllocate(WalSegSz, &SimpleXLogPageRead,
+									&private);
 	if (xlogreader == NULL)
 		pg_fatal("out of memory\n");
 
@@ -170,11 +172,17 @@ findLastCheckpoint(const char *datadir, XLogRecPtr forkptr, int tliIndex,
 	 * header in that case to find the next record.
 	 */
 	if (forkptr % XLOG_BLCKSZ == 0)
-		forkptr += (forkptr % XLogSegSize == 0) ? SizeOfXLogLongPHD : SizeOfXLogShortPHD;
+	{
+		if (XLogSegmentOffset(forkptr, WalSegSz) == 0)
+			forkptr += SizeOfXLogLongPHD;
+		else
+			forkptr += SizeOfXLogShortPHD;
+	}
 
 	private.datadir = datadir;
 	private.tliIndex = tliIndex;
-	xlogreader = XLogReaderAllocate(&SimpleXLogPageRead, &private);
+	xlogreader = XLogReaderAllocate(WalSegSz, &SimpleXLogPageRead,
+									&private);
 	if (xlogreader == NULL)
 		pg_fatal("out of memory\n");
 
@@ -239,21 +247,22 @@ SimpleXLogPageRead(XLogReaderState *xlogreader, XLogRecPtr targetPagePtr,
 	XLogRecPtr	targetSegEnd;
 	XLogSegNo	targetSegNo;
 
-	XLByteToSeg(targetPagePtr, targetSegNo);
-	XLogSegNoOffsetToRecPtr(targetSegNo + 1, 0, targetSegEnd);
-	targetPageOff = targetPagePtr % XLogSegSize;
+	XLByteToSeg(targetPagePtr, targetSegNo, WalSegSz);
+	XLogSegNoOffsetToRecPtr(targetSegNo + 1, 0, targetSegEnd, WalSegSz);
+	targetPageOff = XLogSegmentOffset(targetPagePtr, WalSegSz);
 
 	/*
 	 * See if we need to switch to a new segment because the requested record
 	 * is not in the currently open one.
 	 */
-	if (xlogreadfd >= 0 && !XLByteInSeg(targetPagePtr, xlogreadsegno))
+	if (xlogreadfd >= 0 &&
+		!XLByteInSeg(targetPagePtr, xlogreadsegno, WalSegSz))
 	{
 		close(xlogreadfd);
 		xlogreadfd = -1;
 	}
 
-	XLByteToSeg(targetPagePtr, xlogreadsegno);
+	XLByteToSeg(targetPagePtr, xlogreadsegno, WalSegSz);
 
 	if (xlogreadfd < 0)
 	{
@@ -272,7 +281,8 @@ SimpleXLogPageRead(XLogReaderState *xlogreader, XLogRecPtr targetPagePtr,
 			   targetHistory[private->tliIndex].begin >= targetSegEnd)
 			private->tliIndex--;
 
-		XLogFileName(xlogfname, targetHistory[private->tliIndex].tli, xlogreadsegno);
+		XLogFileName(xlogfname, targetHistory[private->tliIndex].tli,
+					 xlogreadsegno, WalSegSz);
 
 		snprintf(xlogfpath, MAXPGPATH, "%s/" XLOGDIR "/%s", private->datadir, xlogfname);
 

src/bin/pg_rewind/pg_rewind.c

@@ -44,6 +44,7 @@ static ControlFileData ControlFile_target;
 static ControlFileData ControlFile_source;
 
 const char *progname;
+int			WalSegSz;
 
 /* Configuration options */
 char	   *datadir_target = NULL;
@@ -572,8 +573,8 @@ createBackupLabel(XLogRecPtr startpoint, TimeLineID starttli, XLogRecPtr checkpo
 	char		buf[1000];
 	int			len;
 
-	XLByteToSeg(startpoint, startsegno);
-	XLogFileName(xlogfilename, starttli, startsegno);
+	XLByteToSeg(startpoint, startsegno, WalSegSz);
+	XLogFileName(xlogfilename, starttli, startsegno, WalSegSz);
 
 	/*
 	 * Construct backup label file
@@ -631,6 +632,13 @@ digestControlFile(ControlFileData *ControlFile, char *src, size_t size)
 
 	memcpy(ControlFile, src, sizeof(ControlFileData));
 
+	/* set and validate WalSegSz */
+	WalSegSz = ControlFile->xlog_seg_size;
+
+	if (!IsValidWalSegSize(WalSegSz))
+		pg_fatal("WAL segment size must be a power of two between 1MB and 1GB, but the control file specifies %d bytes\n",
+				 WalSegSz);
+
 	/* Additional checks on control file */
 	checkControlFile(ControlFile);
 }

src/bin/pg_rewind/pg_rewind.h

@@ -24,6 +24,7 @@ extern char *connstr_source;
 extern bool debug;
 extern bool showprogress;
 extern bool dry_run;
+extern int	WalSegSz;
 
 /* Target history */
 extern TimeLineHistoryEntry *targetHistory;

src/bin/pg_test_fsync/pg_test_fsync.c

@@ -64,7 +64,7 @@ static const char *progname;
 
 static int	secs_per_test = 5;
 static int	needs_unlink = 0;
-static char full_buf[XLOG_SEG_SIZE],
+static char full_buf[DEFAULT_XLOG_SEG_SIZE],
 		   *buf,
 		   *filename = FSYNC_FILENAME;
 static struct timeval start_t,
@@ -209,7 +209,7 @@ prepare_buf(void)
 	int			ops;
 
 	/* write random data into buffer */
-	for (ops = 0; ops < XLOG_SEG_SIZE; ops++)
+	for (ops = 0; ops < DEFAULT_XLOG_SEG_SIZE; ops++)
 		full_buf[ops] = random();
 
 	buf = (char *) TYPEALIGN(XLOG_BLCKSZ, full_buf);
@@ -226,7 +226,8 @@ test_open(void)
 	if ((tmpfile = open(filename, O_RDWR | O_CREAT, S_IRUSR | S_IWUSR)) == -1)
 		die("could not open output file");
 	needs_unlink = 1;
-	if (write(tmpfile, full_buf, XLOG_SEG_SIZE) != XLOG_SEG_SIZE)
+	if (write(tmpfile, full_buf, DEFAULT_XLOG_SEG_SIZE) !=
+		DEFAULT_XLOG_SEG_SIZE)
 		die("write failed");
 
 	/* fsync now so that dirty buffers don't skew later tests */

src/bin/pg_upgrade/test.sh

@@ -20,7 +20,9 @@ unset MAKELEVEL
 # Run a given "initdb" binary and overlay the regression testing
 # authentication configuration.
 standard_initdb() {
-	"$1" -N
+	# To increase coverage of non-standard segment size without
+	# increase test runtime, run these tests with a lower setting.
+	"$1" -N --wal-segsize 1
 	if [ -n "$TEMP_CONFIG" -a -r "$TEMP_CONFIG" ]
 	then
 		cat "$TEMP_CONFIG" >> "$PGDATA/postgresql.conf"

src/include/access/xlog.h

@@ -94,6 +94,7 @@ extern PGDLLIMPORT XLogRecPtr XactLastCommitEnd;
 extern bool reachedConsistency;
 
 /* these variables are GUC parameters related to XLOG */
+extern int	wal_segment_size;
 extern int	min_wal_size_mb;
 extern int	max_wal_size_mb;
 extern int	wal_keep_segments;

src/include/access/xlog_internal.h

@@ -85,15 +85,27 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
 #define XLogPageHeaderSize(hdr)		\
 	(((hdr)->xlp_info & XLP_LONG_HEADER) ? SizeOfXLogLongPHD : SizeOfXLogShortPHD)
 
-/*
- * The XLOG is split into WAL segments (physical files) of the size indicated
- * by XLOG_SEG_SIZE.
- */
-#define XLogSegSize		((uint32) XLOG_SEG_SIZE)
-#define XLogSegmentsPerXLogId	(UINT64CONST(0x100000000) / XLOG_SEG_SIZE)
+/* wal_segment_size can range from 1MB to 1GB */
+#define WalSegMinSize 1024 * 1024
+#define WalSegMaxSize 1024 * 1024 * 1024
+/* default number of min and max wal segments */
+#define DEFAULT_MIN_WAL_SEGS 5
+#define DEFAULT_MAX_WAL_SEGS 64
+
+/* check that the given size is a valid wal_segment_size */
+#define IsPowerOf2(x) (x > 0 && ((x) & ((x)-1)) == 0)
+#define IsValidWalSegSize(size) \
+	 (IsPowerOf2(size) && \
+	 ((size) >= WalSegMinSize && (size) <= WalSegMaxSize))
+
+#define XLogSegmentsPerXLogId(wal_segsz_bytes)	\
+	(UINT64CONST(0x100000000) / (wal_segsz_bytes))
+
+#define XLogSegNoOffsetToRecPtr(segno, offset, dest, wal_segsz_bytes) \
+		(dest) = (segno) * (wal_segsz_bytes) + (offset)
 
-#define XLogSegNoOffsetToRecPtr(segno, offset, dest) \
-		(dest) = (segno) * XLOG_SEG_SIZE + (offset)
+#define XLogSegmentOffset(xlogptr, wal_segsz_bytes)	\
+	((xlogptr) & ((wal_segsz_bytes) - 1))
 
 /*
  * Compute a segment number from an XLogRecPtr.
@@ -103,11 +115,11 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
  * for deciding which segment to write given a pointer to a record end,
  * for example.
  */
-#define XLByteToSeg(xlrp, logSegNo) \
-	logSegNo = (xlrp) / XLogSegSize
+#define XLByteToSeg(xlrp, logSegNo, wal_segsz_bytes) \
+	logSegNo = (xlrp) / (wal_segsz_bytes)
 
-#define XLByteToPrevSeg(xlrp, logSegNo) \
-	logSegNo = ((xlrp) - 1) / XLogSegSize
+#define XLByteToPrevSeg(xlrp, logSegNo, wal_segsz_bytes) \
+	logSegNo = ((xlrp) - 1) / (wal_segsz_bytes)
 
 /*
  * Is an XLogRecPtr within a particular XLOG segment?
@@ -115,11 +127,11 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
  * For XLByteInSeg, do the computation at face value.  For XLByteInPrevSeg,
  * a boundary byte is taken to be in the previous segment.
  */
-#define XLByteInSeg(xlrp, logSegNo) \
-	(((xlrp) / XLogSegSize) == (logSegNo))
+#define XLByteInSeg(xlrp, logSegNo, wal_segsz_bytes) \
+	(((xlrp) / (wal_segsz_bytes)) == (logSegNo))
 
-#define XLByteInPrevSeg(xlrp, logSegNo) \
-	((((xlrp) - 1) / XLogSegSize) == (logSegNo))
+#define XLByteInPrevSeg(xlrp, logSegNo, wal_segsz_bytes) \
+	((((xlrp) - 1) / (wal_segsz_bytes)) == (logSegNo))
 
 /* Check if an XLogRecPtr value is in a plausible range */
 #define XRecOffIsValid(xlrp) \
@@ -140,10 +152,10 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
 /* Length of XLog file name */
 #define XLOG_FNAME_LEN	   24
 
-#define XLogFileName(fname, tli, logSegNo)	\
+#define XLogFileName(fname, tli, logSegNo, wal_segsz_bytes)	\
 	snprintf(fname, MAXFNAMELEN, "%08X%08X%08X", tli,		\
-			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId), \
-			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId))
+			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId(wal_segsz_bytes)))
 
 #define XLogFileNameById(fname, tli, log, seg)	\
 	snprintf(fname, MAXFNAMELEN, "%08X%08X%08X", tli, log, seg)
@@ -162,18 +174,18 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
 	 strspn(fname, "0123456789ABCDEF") == XLOG_FNAME_LEN &&		\
 	 strcmp((fname) + XLOG_FNAME_LEN, ".partial") == 0)
 
-#define XLogFromFileName(fname, tli, logSegNo)	\
+#define XLogFromFileName(fname, tli, logSegNo, wal_segsz_bytes)	\
 	do {												\
 		uint32 log;										\
 		uint32 seg;										\
 		sscanf(fname, "%08X%08X%08X", tli, &log, &seg); \
-		*logSegNo = (uint64) log * XLogSegmentsPerXLogId + seg; \
+		*logSegNo = (uint64) log * XLogSegmentsPerXLogId(wal_segsz_bytes) + seg; \
 	} while (0)
 
-#define XLogFilePath(path, tli, logSegNo)	\
-	snprintf(path, MAXPGPATH, XLOGDIR "/%08X%08X%08X", tli,				\
-			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId),				\
-			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId))
+#define XLogFilePath(path, tli, logSegNo, wal_segsz_bytes)	\
+	snprintf(path, MAXPGPATH, XLOGDIR "/%08X%08X%08X", tli,	\
+			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId(wal_segsz_bytes)))
 
 #define TLHistoryFileName(fname, tli)	\
 	snprintf(fname, MAXFNAMELEN, "%08X.history", tli)
@@ -189,20 +201,22 @@ typedef XLogLongPageHeaderData *XLogLongPageHeader;
 #define StatusFilePath(path, xlog, suffix)	\
 	snprintf(path, MAXPGPATH, XLOGDIR "/archive_status/%s%s", xlog, suffix)
 
-#define BackupHistoryFileName(fname, tli, logSegNo, offset) \
+#define BackupHistoryFileName(fname, tli, logSegNo, startpoint, wal_segsz_bytes) \
 	snprintf(fname, MAXFNAMELEN, "%08X%08X%08X.%08X.backup", tli, \
-			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId),		  \
-			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId), offset)
+			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) (XLogSegmentOffset(startpoint, wal_segsz_bytes)))
 
 #define IsBackupHistoryFileName(fname) \
 	(strlen(fname) > XLOG_FNAME_LEN && \
 	 strspn(fname, "0123456789ABCDEF") == XLOG_FNAME_LEN && \
 	 strcmp((fname) + strlen(fname) - strlen(".backup"), ".backup") == 0)
 
-#define BackupHistoryFilePath(path, tli, logSegNo, offset)	\
+#define BackupHistoryFilePath(path, tli, logSegNo, startpoint, wal_segsz_bytes)	\
 	snprintf(path, MAXPGPATH, XLOGDIR "/%08X%08X%08X.%08X.backup", tli, \
-			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId), \
-			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId), offset)
+			 (uint32) ((logSegNo) / XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) ((logSegNo) % XLogSegmentsPerXLogId(wal_segsz_bytes)), \
+			 (uint32) (XLogSegmentOffset((startpoint), wal_segsz_bytes)))
 
 /*
  * Information logged when we detect a change in one of the parameters

src/include/access/xlogreader.h

@@ -73,6 +73,11 @@ struct XLogReaderState
 	 * ----------------------------------------
 	 */
 
+	/*
+	 * Segment size of the to-be-parsed data (mandatory).
+	 */
+	int			wal_segment_size;
+
 	/*
 	 * Data input callback (mandatory).
 	 *
@@ -189,7 +194,8 @@ struct XLogReaderState
 };
 
 /* Get a new XLogReader */
-extern XLogReaderState *XLogReaderAllocate(XLogPageReadCB pagereadfunc,
+extern XLogReaderState *XLogReaderAllocate(int wal_segment_size,
+				   XLogPageReadCB pagereadfunc,
 				   void *private_data);
 
 /* Free an XLogReader */

src/include/catalog/pg_control.h

@@ -21,7 +21,7 @@
 
 
 /* Version identifier for this pg_control format */
-#define PG_CONTROL_VERSION	1002
+#define PG_CONTROL_VERSION	1003
 
 /* Nonce key length, see below */
 #define MOCK_AUTH_NONCE_LEN		32

src/include/pg_config.h.in

@@ -895,11 +895,6 @@
    */
 #undef XLOG_BLCKSZ
 
-/* XLOG_SEG_SIZE is the size of a single WAL file. This must be a power of 2
-   and larger than XLOG_BLCKSZ (preferably, a great deal larger than
-   XLOG_BLCKSZ). Changing XLOG_SEG_SIZE requires an initdb. */
-#undef XLOG_SEG_SIZE
-
 
 
 /* Number of bits in a file offset, on hosts where this is settable. */

src/include/pg_config_manual.h

@@ -13,6 +13,12 @@
  *------------------------------------------------------------------------
  */
 
+/*
+ * This is default value for wal_segment_size to be used at intidb when run
+ * without --walsegsize option. Must be a valid segment size.
+ */
+#define DEFAULT_XLOG_SEG_SIZE	(16*1024*1024)
+
 /*
  * Maximum length for identifiers (e.g. table names, column names,
  * function names).  Names actually are limited to one less byte than this,

src/tools/msvc/Solution.pm

@@ -179,8 +179,6 @@ s{PG_VERSION_STR "[^"]+"}{PG_VERSION_STR "PostgreSQL $self->{strver}$extraver, c
 		  1024, "\n";
 		print $o "#define XLOG_BLCKSZ ",
 		  1024 * $self->{options}->{wal_blocksize}, "\n";
-		print $o "#define XLOG_SEG_SIZE (", $self->{options}->{wal_segsize},
-		  " * 1024 * 1024)\n";
 
 		if ($self->{options}->{float4byval})
 		{