diff --git a/Makefile b/Makefile
index 902a976779..cef149dec9 100644
--- a/Makefile
+++ b/Makefile
@@ -1583,7 +1583,10 @@ u-boot.spr: spl/u-boot-spl.img u-boot.img FORCE
 
 ifneq ($(CONFIG_ARCH_SOCFPGA),)
 quiet_cmd_gensplx4 = GENSPLX4 $@
-cmd_gensplx4 = cat	spl/u-boot-spl.sfp spl/u-boot-spl.sfp	\
+cmd_gensplx4 = $(OBJCOPY) -I binary -O binary --gap-fill=0x0		\
+			--pad-to=$(CONFIG_SPL_PAD_TO)			\
+			spl/u-boot-spl.sfp spl/u-boot-spl.sfp &&        \
+		cat	spl/u-boot-spl.sfp spl/u-boot-spl.sfp		\
 			spl/u-boot-spl.sfp spl/u-boot-spl.sfp > $@ || { rm -f $@; false; }
 spl/u-boot-splx4.sfp: spl/u-boot-spl.sfp FORCE
 	$(call if_changed,gensplx4)
diff --git a/tools/socfpgaimage.c b/tools/socfpgaimage.c
index 5808b383e9..eba812fec9 100644
--- a/tools/socfpgaimage.c
+++ b/tools/socfpgaimage.c
@@ -61,6 +61,7 @@
 
 #define HEADER_OFFSET	0x40
 #define VALIDATION_WORD	0x31305341
+#define IMAGE_ALIGN	16
 
 /* Minimum and default entry point offset */
 #define ENTRY_POINT_OFFSET	0x14
@@ -97,7 +98,7 @@ static unsigned int sfp_hdr_size(uint8_t ver)
 	return 0;
 }
 
-static unsigned int sfp_pad_size(uint8_t ver)
+static unsigned int sfp_max_size(uint8_t ver)
 {
 	if (ver == 0)
 		return sizeof(buffer_v0);
@@ -106,6 +107,12 @@ static unsigned int sfp_pad_size(uint8_t ver)
 	return 0;
 }
 
+static unsigned int sfp_aligned_len(uint32_t size)
+{
+	/* Add 4 bytes for CRC and align to 16 bytes */
+	return ALIGN(size + sizeof(uint32_t), IMAGE_ALIGN);
+}
+
 /*
  * The header checksum is just a very simple checksum over
  * the header area.
@@ -208,22 +215,24 @@ static int sfp_sign_buffer(uint8_t *buf, uint8_t ver, uint8_t flags,
 			   struct image_tool_params *params)
 {
 	uint32_t calc_crc;
+	uint32_t crc_off;
 
 	/* Align the length up */
-	len = ALIGN(len, 4);
+	len = sfp_aligned_len(len);
 
-	/* Build header, adding 4 bytes to length to hold the CRC32. */
-	sfp_build_header(buf + HEADER_OFFSET, ver, flags, len + 4, params);
+	/* Build header */
+	sfp_build_header(buf + HEADER_OFFSET, ver, flags, len, params);
 
 	/* Calculate and apply the CRC */
-	calc_crc = ~pbl_crc32(0, (char *)buf, len);
+	crc_off = len - sizeof(uint32_t); /* at last 4 bytes of image */
+	calc_crc = ~pbl_crc32(0, (char *)buf, crc_off);
 
-	*((uint32_t *)(buf + len)) = cpu_to_le32(calc_crc);
+	*((uint32_t *)(buf + crc_off)) = cpu_to_le32(calc_crc);
 
 	if (!pad_64k)
 		return len + 4;
 
-	return sfp_pad_size(ver);
+	return sfp_max_size(ver);
 }
 
 /* Verify that the buffer looks sane */
@@ -240,7 +249,7 @@ static int sfp_verify_buffer(const uint8_t *buf)
 		return -1;
 	}
 
-	if (len < HEADER_OFFSET || len > sfp_pad_size(ver)) {
+	if (len < HEADER_OFFSET || len > sfp_max_size(ver)) {
 		debug("Invalid header length (%i)\n", len);
 		return -1;
 	}
@@ -367,19 +376,25 @@ static int socfpgaimage_check_image_types_v1(uint8_t type)
  * To work in with the mkimage framework, we do some ugly stuff...
  *
  * First, socfpgaimage_vrec_header() is called.
- * We prepend a fake header big enough to make the file sfp_pad_size().
+ * We prepend a fake header big enough to include crc32 and align image to 16
+ * bytes.
  * This gives us enough space to do what we want later.
  *
  * Next, socfpgaimage_set_header() is called.
  * We fix up the buffer by moving the image to the start of the buffer.
- * We now have some room to do what we need (add CRC and padding).
+ * We now have some room to do what we need (add CRC).
  */
 
 static int data_size;
 
 static int sfp_fake_header_size(unsigned int size, uint8_t ver)
 {
-	return sfp_pad_size(ver) - size;
+	unsigned int align_size;
+
+	align_size = sfp_aligned_len(size);
+
+	/* extra bytes needed */
+	return align_size - size;
 }
 
 static int sfp_vrec_header(struct image_tool_params *params,
@@ -389,7 +404,7 @@ static int sfp_vrec_header(struct image_tool_params *params,
 
 	if (params->datafile &&
 	    stat(params->datafile, &sbuf) == 0 &&
-	    sbuf.st_size <= (sfp_pad_size(ver) - sizeof(uint32_t))) {
+	    sbuf.st_size <= (sfp_max_size(ver) - sizeof(uint32_t))) {
 		data_size = sbuf.st_size;
 		tparams->header_size = sfp_fake_header_size(data_size, ver);
 	}
@@ -417,7 +432,7 @@ static void sfp_set_header(void *ptr, unsigned char ver,
 	/*
 	 * This function is called after vrec_header() has been called.
 	 * At this stage we have the sfp_fake_header_size() dummy bytes
-	 * followed by data_size image bytes. Total = sfp_pad_size().
+	 * followed by data_size image bytes.
 	 * We need to fix the buffer by moving the image bytes back to
 	 * the beginning of the buffer, then actually do the signing stuff...
 	 */