Skip to contentSkip to conversation

Revert "tree: imx: remove old fit generator script"

PATCH In review Tom Rini 5 participants 20 messages Jan 6, 2022 – Jan 7, 2022

Subsystems (4)
lore
Share & tools
lore

Tom Rini

author
#0 lore raw

PATCH 1/1 Revert "tree: imx: remove old fit generator script"

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series that allows for binman to be used to generate the final images, and also not break CI, and then also merge all of the series that convert currently broken platforms to use binman instead. So, bring back this script now and remove it again for real after the release.

---
 Makefile                             |   3 +
 arch/arm/mach-imx/mkimage_fit_atf.sh | 143 +++++++++++++++++++++++++++
 2 files changed, 146 insertions(+)
 create mode 100755 arch/arm/mach-imx/mkimage_fit_atf.sh
Makefile +3 -0
index ae9bfab91acf..0c478645c3a7 100644
@@ -1343,6 +1343,9 @@ $(U_BOOT_ITS): $(subst ",,$(CONFIG_SPL_FIT_SOURCE))
 else
 ifneq ($(CONFIG_USE_SPL_FIT_GENERATOR),)
 U_BOOT_ITS := u-boot.its
+ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-imx/mkimage_fit_atf.sh")
+U_BOOT_ITS_DEPS += u-boot-nodtb.bin
+endif
 ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-rockchip/make_fit_atf.py")
 U_BOOT_ITS_DEPS += u-boot
 endif
arch/arm/mach-imx/mkimage_fit_atf.sh new file +143 -0

143 lines not on this page · raw message

Signature · 1 line
2.25.1
#1 lore raw

Hello Tom,

Quoted 7 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator script"

Quoted 7 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series that allows for binman to be used to generate the final images, and also not break CI, and then also merge all of the series that convert currently broken platforms to use binman instead. So, bring back this script now and remove it again for real after the release.

Please note that this might not work, as the FIT generator script would generate ITS with '@' symbols which are not compatible with mkimage due to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Quoted 77 lines
---
 Makefile                             |   3 +
 arch/arm/mach-imx/mkimage_fit_atf.sh | 143 +++++++++++++++++++++++++++
 2 files changed, 146 insertions(+)
 create mode 100755 arch/arm/mach-imx/mkimage_fit_atf.sh
Makefile +3 -0
index ae9bfab91acf..0c478645c3a7 100644
@@ -1343,6 +1343,9 @@ $(U_BOOT_ITS): $(subst ",,$(CONFIG_SPL_FIT_SOURCE))
 else
 ifneq ($(CONFIG_USE_SPL_FIT_GENERATOR),)
 U_BOOT_ITS := u-boot.its
+ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-imx/mkimage_fit_atf.sh")
+U_BOOT_ITS_DEPS += u-boot-nodtb.bin
+endif
 ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-rockchip/make_fit_atf.py")
 U_BOOT_ITS_DEPS += u-boot
 endif
arch/arm/mach- renamed from arch/arm/mach-imx/mkimage_fit_atf.sh +0 -0

imx/mkimage_fit_atf.sh
new file mode 100755
index 000000000000..2a17968794c1

arch/arm/mach-imx/mkimage_fit_atf.sh new file +27 -0
@@ -0,0 +1,143 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0+
+#
+# script to generate FIT image source for i.MX8MQ boards with
+# ARM Trusted Firmware and multiple device trees (given on the command line)
+#
+# usage: $0 <dt_name> [<dt_name> [<dt_name] ...]
+
+[ -z "$BL31" ] && BL31="bl31.bin"
+[ -z "$TEE_LOAD_ADDR" ] && TEE_LOAD_ADDR="0xfe000000"
+[ -z "$ATF_LOAD_ADDR" ] && ATF_LOAD_ADDR="0x00910000"
+[ -z "$BL33_LOAD_ADDR" ] && BL33_LOAD_ADDR="0x40200000"
+
+if [ ! -f $BL31 ]; then
+	echo "ERROR: BL31 file $BL31 NOT found" >&2
+	exit 0
+else
+	echo "$BL31 size: " >&2
+	stat -c %s $BL31 >&2
+fi
+
+BL32="tee.bin"
+
+if [ ! -f $BL32 ]; then
+	BL32=/dev/null
+else
+	echo "Building with TEE support, make sure your $BL31 is compiled with

spd. If you do not want tee, please delete $BL31" >&2

arch/arm/mach-imx/mkimage_fit_atf.sh new file +19 -0
+	echo "$BL32 size: " >&2
+	stat -c %s $BL32 >&2
+fi
+
+BL33="u-boot-nodtb.bin"
+
+if [ ! -f $BL33 ]; then
+	echo "ERROR: $BL33 file NOT found" >&2
+	exit 0
+else
+	echo "u-boot-nodtb.bin size: " >&2
+	stat -c %s u-boot-nodtb.bin >&2
+fi
+
+for dtname in $*
+do
+	echo "$dtname size: " >&2
+	stat -c %s $dtname >&2
+done

arch/arm/mach-imx/mkimage_fit_atf.sh line 56

Quoted 10 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +10 -0
+
+
+cat << __HEADER_EOF
+/dts-v1/;
+
+/ {
+	description = "Configuration to load ATF before U-Boot";
+
+	images {
+		uboot@1 {

This (and all other similar places) would be rejected by mkimage... :(

Quoted 88 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +87 -1
+			description = "U-Boot (64-bit)";
+			os = "u-boot";
+			data = /incbin/("$BL33");
+			type = "standalone";
+			arch = "arm64";
+			compression = "none";
+			load = <$BL33_LOAD_ADDR>;
+		};
+__HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+	cat << __FDT_IMAGE_EOF
+		fdt@$cnt {
+			description = "$(basename $dtname .dtb)";
+			data = /incbin/("$dtname");
+			type = "flat_dt";
+			compression = "none";
+		};
+__FDT_IMAGE_EOF
+cnt=$((cnt+1))
+done
+
+cat << __HEADER_EOF
+		atf@1 {
+			description = "ARM Trusted Firmware";
+			os = "arm-trusted-firmware";
+			data = /incbin/("$BL31");
+			type = "firmware";
+			arch = "arm64";
+			compression = "none";
+			load = <$ATF_LOAD_ADDR>;
+			entry = <$ATF_LOAD_ADDR>;
+		};
+__HEADER_EOF
+
+if [ -f $BL32 ]; then
+cat << __HEADER_EOF
+		tee@1 {
+			description = "TEE firmware";
+			data = /incbin/("$BL32");
+			type = "firmware";
+			arch = "arm64";
+			compression = "none";
+			load = <$TEE_LOAD_ADDR>;
+			entry = <$TEE_LOAD_ADDR>;
+		};
+__HEADER_EOF
+fi
+
+cat << __CONF_HEADER_EOF
+	};
+	configurations {
+		default = "config@1";
+
+__CONF_HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+if [ -f $BL32 ]; then
+cat << __CONF_SECTION_EOF
+		config@$cnt {
+			description = "$(basename $dtname .dtb)";
+			firmware = "uboot@1";
+			loadables = "atf@1", "tee@1";
+			fdt = "fdt@$cnt";
+		};
+__CONF_SECTION_EOF
+else
+cat << __CONF_SECTION1_EOF
+		config@$cnt {
+			description = "$(basename $dtname .dtb)";
+			firmware = "uboot@1";
+			loadables = "atf@1";
+			fdt = "fdt@$cnt";
+		};
+__CONF_SECTION1_EOF
+fi
+cnt=$((cnt+1))
+done
+
+cat << __ITS_EOF
+	};
+};
+__ITS_EOF
--
Quoted 1 line

2.25.1

-- andrey

#2 lore raw
Quoted 18 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 7 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator script"

Quoted 7 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series that allows for binman to be used to generate the final images, and also not break CI, and then also merge all of the series that convert currently broken platforms to use binman instead. So, bring back this script now and remove it again for real after the release.

Quoted 4 lines

Please note that this might not work, as the FIT generator script would generate ITS with '@' symbols which are not compatible with mkimage due to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property
u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property
u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property
u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property
u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property
u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property
u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property
u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit code 1 Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or directory arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":
configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 183 lines
Quoted 78 lines
---
 Makefile                             |   3 +
 arch/arm/mach-imx/mkimage_fit_atf.sh | 143 +++++++++++++++++++++++++++
 2 files changed, 146 insertions(+)
 create mode 100755 arch/arm/mach-imx/mkimage_fit_atf.sh
Makefile +3 -0
index ae9bfab91acf..0c478645c3a7 100644
@@ -1343,6 +1343,9 @@ $(U_BOOT_ITS): $(subst ",,$(CONFIG_SPL_FIT_SOURCE))
 else
 ifneq ($(CONFIG_USE_SPL_FIT_GENERATOR),)
 U_BOOT_ITS := u-boot.its
+ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-imx/mkimage_fit_atf.sh")
+U_BOOT_ITS_DEPS += u-boot-nodtb.bin
+endif
 ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-rockchip/make_fit_atf.py")
 U_BOOT_ITS_DEPS += u-boot
 endif
arch/arm/mach- renamed from arch/arm/mach-imx/mkimage_fit_atf.sh +0 -0

imx/mkimage_fit_atf.sh
new file mode 100755
index 000000000000..2a17968794c1

arch/arm/mach-imx/mkimage_fit_atf.sh new file +27 -0
@@ -0,0 +1,143 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0+
+#
+# script to generate FIT image source for i.MX8MQ boards with
+# ARM Trusted Firmware and multiple device trees (given on the command line)
+#
+# usage: $0 <dt_name> [<dt_name> [<dt_name] ...]
+
+[ -z "$BL31" ] && BL31="bl31.bin"
+[ -z "$TEE_LOAD_ADDR" ] && TEE_LOAD_ADDR="0xfe000000"
+[ -z "$ATF_LOAD_ADDR" ] && ATF_LOAD_ADDR="0x00910000"
+[ -z "$BL33_LOAD_ADDR" ] && BL33_LOAD_ADDR="0x40200000"
+
+if [ ! -f $BL31 ]; then
+     echo "ERROR: BL31 file $BL31 NOT found" >&2
+     exit 0
+else
+     echo "$BL31 size: " >&2
+     stat -c %s $BL31 >&2
+fi
+
+BL32="tee.bin"
+
+if [ ! -f $BL32 ]; then
+     BL32=/dev/null
+else
+     echo "Building with TEE support, make sure your $BL31 is compiled with

spd. If you do not want tee, please delete $BL31" >&2

arch/arm/mach-imx/mkimage_fit_atf.sh new file +19 -0
+     echo "$BL32 size: " >&2
+     stat -c %s $BL32 >&2
+fi
+
+BL33="u-boot-nodtb.bin"
+
+if [ ! -f $BL33 ]; then
+     echo "ERROR: $BL33 file NOT found" >&2
+     exit 0
+else
+     echo "u-boot-nodtb.bin size: " >&2
+     stat -c %s u-boot-nodtb.bin >&2
+fi
+
+for dtname in $*
+do
+     echo "$dtname size: " >&2
+     stat -c %s $dtname >&2
+done

arch/arm/mach-imx/mkimage_fit_atf.sh line 56

Quoted 10 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +10 -0
+
+
+cat << __HEADER_EOF
+/dts-v1/;
+
+/ {
+     description = "Configuration to load ATF before U-Boot";
+
+     images {
+             uboot@1 {

This (and all other similar places) would be rejected by mkimage... :(

Quoted 88 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +87 -1
+                     description = "U-Boot (64-bit)";
+                     os = "u-boot";
+                     data = /incbin/("$BL33");
+                     type = "standalone";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$BL33_LOAD_ADDR>;
+             };
+__HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+     cat << __FDT_IMAGE_EOF
+             fdt@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     data = /incbin/("$dtname");
+                     type = "flat_dt";
+                     compression = "none";
+             };
+__FDT_IMAGE_EOF
+cnt=$((cnt+1))
+done
+
+cat << __HEADER_EOF
+             atf@1 {
+                     description = "ARM Trusted Firmware";
+                     os = "arm-trusted-firmware";
+                     data = /incbin/("$BL31");
+                     type = "firmware";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$ATF_LOAD_ADDR>;
+                     entry = <$ATF_LOAD_ADDR>;
+             };
+__HEADER_EOF
+
+if [ -f $BL32 ]; then
+cat << __HEADER_EOF
+             tee@1 {
+                     description = "TEE firmware";
+                     data = /incbin/("$BL32");
+                     type = "firmware";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$TEE_LOAD_ADDR>;
+                     entry = <$TEE_LOAD_ADDR>;
+             };
+__HEADER_EOF
+fi
+
+cat << __CONF_HEADER_EOF
+     };
+     configurations {
+             default = "config@1";
+
+__CONF_HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+if [ -f $BL32 ]; then
+cat << __CONF_SECTION_EOF
+             config@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     firmware = "uboot@1";
+                     loadables = "atf@1", "tee@1";
+                     fdt = "fdt@$cnt";
+             };
+__CONF_SECTION_EOF
+else
+cat << __CONF_SECTION1_EOF
+             config@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     firmware = "uboot@1";
+                     loadables = "atf@1";
+                     fdt = "fdt@$cnt";
+             };
+__CONF_SECTION1_EOF
+fi
+cnt=$((cnt+1))
+done
+
+cat << __ITS_EOF
+     };
+};
+__ITS_EOF
--
Quoted 1 line

2.25.1

-- andrey

#3 lore raw
Quoted 104 lines · Tim Harvey
Quoted 18 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 7 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator script"

Quoted 7 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series that allows for binman to be used to generate the final images, and also not break CI, and then also merge all of the series that convert currently broken platforms to use binman instead. So, bring back this script now and remove it again for real after the release.

Quoted 4 lines

Please note that this might not work, as the FIT generator script would generate ITS with '@' symbols which are not compatible with mkimage due to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property
u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property
u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property
u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property
u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property
u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property
u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property
u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit code 1 Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or directory arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":
configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 6 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool. Is it a show-stopper that it does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Best regards,

Tim

Quoted 183 lines
Quoted 183 lines
Quoted 78 lines
---
 Makefile                             |   3 +
 arch/arm/mach-imx/mkimage_fit_atf.sh | 143 +++++++++++++++++++++++++++
 2 files changed, 146 insertions(+)
 create mode 100755 arch/arm/mach-imx/mkimage_fit_atf.sh
Makefile +3 -0
index ae9bfab91acf..0c478645c3a7 100644
@@ -1343,6 +1343,9 @@ $(U_BOOT_ITS): $(subst ",,$(CONFIG_SPL_FIT_SOURCE))
 else
 ifneq ($(CONFIG_USE_SPL_FIT_GENERATOR),)
 U_BOOT_ITS := u-boot.its
+ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-imx/mkimage_fit_atf.sh")
+U_BOOT_ITS_DEPS += u-boot-nodtb.bin
+endif
 ifeq ($(CONFIG_SPL_FIT_GENERATOR),"arch/arm/mach-rockchip/make_fit_atf.py")
 U_BOOT_ITS_DEPS += u-boot
 endif
arch/arm/mach- renamed from arch/arm/mach-imx/mkimage_fit_atf.sh +0 -0

imx/mkimage_fit_atf.sh
new file mode 100755
index 000000000000..2a17968794c1

arch/arm/mach-imx/mkimage_fit_atf.sh new file +27 -0
@@ -0,0 +1,143 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0+
+#
+# script to generate FIT image source for i.MX8MQ boards with
+# ARM Trusted Firmware and multiple device trees (given on the command line)
+#
+# usage: $0 <dt_name> [<dt_name> [<dt_name] ...]
+
+[ -z "$BL31" ] && BL31="bl31.bin"
+[ -z "$TEE_LOAD_ADDR" ] && TEE_LOAD_ADDR="0xfe000000"
+[ -z "$ATF_LOAD_ADDR" ] && ATF_LOAD_ADDR="0x00910000"
+[ -z "$BL33_LOAD_ADDR" ] && BL33_LOAD_ADDR="0x40200000"
+
+if [ ! -f $BL31 ]; then
+     echo "ERROR: BL31 file $BL31 NOT found" >&2
+     exit 0
+else
+     echo "$BL31 size: " >&2
+     stat -c %s $BL31 >&2
+fi
+
+BL32="tee.bin"
+
+if [ ! -f $BL32 ]; then
+     BL32=/dev/null
+else
+     echo "Building with TEE support, make sure your $BL31 is compiled with

spd. If you do not want tee, please delete $BL31" >&2

arch/arm/mach-imx/mkimage_fit_atf.sh new file +19 -0
+     echo "$BL32 size: " >&2
+     stat -c %s $BL32 >&2
+fi
+
+BL33="u-boot-nodtb.bin"
+
+if [ ! -f $BL33 ]; then
+     echo "ERROR: $BL33 file NOT found" >&2
+     exit 0
+else
+     echo "u-boot-nodtb.bin size: " >&2
+     stat -c %s u-boot-nodtb.bin >&2
+fi
+
+for dtname in $*
+do
+     echo "$dtname size: " >&2
+     stat -c %s $dtname >&2
+done

arch/arm/mach-imx/mkimage_fit_atf.sh line 56

Quoted 10 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +10 -0
+
+
+cat << __HEADER_EOF
+/dts-v1/;
+
+/ {
+     description = "Configuration to load ATF before U-Boot";
+
+     images {
+             uboot@1 {

This (and all other similar places) would be rejected by mkimage... :(

Quoted 88 lines
arch/arm/mach-imx/mkimage_fit_atf.sh new file +87 -1
+                     description = "U-Boot (64-bit)";
+                     os = "u-boot";
+                     data = /incbin/("$BL33");
+                     type = "standalone";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$BL33_LOAD_ADDR>;
+             };
+__HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+     cat << __FDT_IMAGE_EOF
+             fdt@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     data = /incbin/("$dtname");
+                     type = "flat_dt";
+                     compression = "none";
+             };
+__FDT_IMAGE_EOF
+cnt=$((cnt+1))
+done
+
+cat << __HEADER_EOF
+             atf@1 {
+                     description = "ARM Trusted Firmware";
+                     os = "arm-trusted-firmware";
+                     data = /incbin/("$BL31");
+                     type = "firmware";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$ATF_LOAD_ADDR>;
+                     entry = <$ATF_LOAD_ADDR>;
+             };
+__HEADER_EOF
+
+if [ -f $BL32 ]; then
+cat << __HEADER_EOF
+             tee@1 {
+                     description = "TEE firmware";
+                     data = /incbin/("$BL32");
+                     type = "firmware";
+                     arch = "arm64";
+                     compression = "none";
+                     load = <$TEE_LOAD_ADDR>;
+                     entry = <$TEE_LOAD_ADDR>;
+             };
+__HEADER_EOF
+fi
+
+cat << __CONF_HEADER_EOF
+     };
+     configurations {
+             default = "config@1";
+
+__CONF_HEADER_EOF
+
+cnt=1
+for dtname in $*
+do
+if [ -f $BL32 ]; then
+cat << __CONF_SECTION_EOF
+             config@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     firmware = "uboot@1";
+                     loadables = "atf@1", "tee@1";
+                     fdt = "fdt@$cnt";
+             };
+__CONF_SECTION_EOF
+else
+cat << __CONF_SECTION1_EOF
+             config@$cnt {
+                     description = "$(basename $dtname .dtb)";
+                     firmware = "uboot@1";
+                     loadables = "atf@1";
+                     fdt = "fdt@$cnt";
+             };
+__CONF_SECTION1_EOF
+fi
+cnt=$((cnt+1))
+done
+
+cat << __ITS_EOF
+     };
+};
+__ITS_EOF
--
Quoted 1 line

2.25.1

-- andrey

author
#4 lore raw
Quoted 112 lines · Tim Harvey
Quoted 104 lines · Tim Harvey
Quoted 18 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 7 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator script"

Quoted 7 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series that allows for binman to be used to generate the final images, and also not break CI, and then also merge all of the series that convert currently broken platforms to use binman instead. So, bring back this script now and remove it again for real after the release.

Quoted 4 lines

Please note that this might not work, as the FIT generator script would generate ITS with '@' symbols which are not compatible with mkimage due to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property
u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property
u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property
u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property
u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property
u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property
u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property
u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit code 1 Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or directory arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":
configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh" configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 6 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 3 lines

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch.
Everything is currently visible at
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is non-functional: blob-ext@1 blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is non-functional: blob-ext +(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:
aarch64: + imx8mn_beacon_2g
+(imx8mn_beacon_2g) ===================== WARNING ====================== +(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please migrate +(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of +(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ==================================================== +(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is non-functional: blob- ext@1 blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl .cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage': spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1 +(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

and that's the problem to fix still, that ddr bin being faked out, in CI. I think I see how to do that however, and I'm going to test that out locally and then shoot this at CI and see what happens.

Signature · 1 line
Tom
#5 lore raw
Quoted 176 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Review

Quoted 3 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

adam

Quoted 8 lines

and that's the problem to fix still, that ddr bin being faked out, in CI. I think I see how to do that however, and I'm going to test that out locally and then shoot this at CI and see what happens.

--
Tom

author
#6 lore raw
Quoted 177 lines · Adam Ford
Quoted 175 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful. As best I can tell there's still something missing with making
fake blobs link and not fail. For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

And that last error is the real problem.

Signature · 1 line
Tom
#7 lore raw
Quoted 229 lines · Tom Rini
Quoted 6 lines · Adam Ford
Quoted 4 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 1 line · wrote:
Quoted 1 line
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom

Rini

Quoted 9 lines
Quoted 9 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the

series

Quoted 4 lines
Quoted 4 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

Quoted 8 lines
Quoted 8 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with

mkimage

Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be complemented with the

fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1

:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg

property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image support with

exit

Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I

stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to binman for

this

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html

but I

Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines

don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin'

failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved

the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also don't know

what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not

get

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman

conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1] and as far

as I

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to

build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman

[3]

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines

and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 5 lines
Quoted 5 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 3 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin

target

Quoted 9 lines
Quoted 9 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 10 lines
Quoted 10 lines

and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid
Quoted 3 lines
Quoted 3 lines

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 5 lines

Review

Quoted 5 lines

+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please migrate

(quoted diff) +3 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and

is

Quoted 5 lines
Quoted 5 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':

Quoted 8 lines

Review

Quoted 6 lines

spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just

grab

Quoted 2 lines

the binman updates and apply it to master and fix it from that starting point?

Quoted 15 lines

So, yes, grabbing the series from the above link and applying it on top of master, and making builds work with fake binaries would be very helpful. As best I can tell there's still something missing with making fake blobs link and not fail. For example, imx8mm_venice is part of the series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is
non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional:
atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

$ cat imx-boot.map
ImagePos    Offset      Size  Name
00000000  00000000  0011d778  main-section
00000000   00000000  00039000  spl
00057c00   00057c00  000c5b78  uboot
$ cat spl.map
ImagePos    Offset      Size  Name
00000000  00000000  00039000  main-section
00000000   00000000  00039000  mkimage
$ cat u-boot-spl-ddr.map
ImagePos    Offset      Size  Name
00000000  00000000  00036f5c  main-section
00000000   00000000  0001ef5c  u-boot-spl
00000000    00000000  0001dc40  u-boot-spl-nodtb
0001dc40    0001dc40  0000131c  u-boot-spl-dtb
0001ef5c   0001ef5c  00008000  1d-imem
00026f5c   00026f5c  00004000  1d_dmem
0002af5c   0002af5c  00008000  2d_imem
00032f5c   00032f5c  00004000  2d_dmem

Tim,

Can you try deleting the &binman node from your mx8mm-venice-u-boot.dtsi file and see if the stock binman node in 8mm-u-boot-dtsi works for you? The new default one does for me on my Beacon 8mm board.

Tom,

I will submit another patch for the imx8mn_beacon boards to fix the build error on the 2g version. I forgot to update the second defconfig file when
I pushed the patch for the first one. I have it working, but I need to
clean it up.

adam

Quoted 5 lines

And that last error is the real problem.

--
Tom

author
#8 lore raw
Quoted 281 lines · Adam Ford
Quoted 229 lines · Tom Rini
Quoted 6 lines · Adam Ford
Quoted 4 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 1 line · wrote:
Quoted 1 line
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom

Rini

Quoted 9 lines
Quoted 9 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the

series

Quoted 4 lines
Quoted 4 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

Quoted 8 lines
Quoted 8 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with

mkimage

Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be complemented with the

fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1

:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg

property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image support with

exit

Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I

stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to binman for

this

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html

but I

Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines

don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin'

failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved

the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also don't know

what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not

get

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman

conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1] and as far

as I

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to

build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman

[3]

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines

and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 5 lines
Quoted 5 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 3 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin

target

Quoted 9 lines
Quoted 9 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 10 lines
Quoted 10 lines

and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid
Quoted 3 lines
Quoted 3 lines

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 5 lines

Review

Quoted 5 lines

+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please migrate

(quoted diff) +3 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and

is

Quoted 5 lines
Quoted 5 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':

Quoted 8 lines

Review

Quoted 6 lines

spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just

grab

Quoted 2 lines

the binman updates and apply it to master and fix it from that starting point?

Quoted 15 lines

So, yes, grabbing the series from the above link and applying it on top of master, and making builds work with fake binaries would be very helpful. As best I can tell there's still something missing with making fake blobs link and not fail. For example, imx8mm_venice is part of the series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is
non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional:
atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

$ cat imx-boot.map
ImagePos    Offset      Size  Name
00000000  00000000  0011d778  main-section
00000000   00000000  00039000  spl
00057c00   00057c00  000c5b78  uboot
$ cat spl.map
ImagePos    Offset      Size  Name
00000000  00000000  00039000  main-section
00000000   00000000  00039000  mkimage
$ cat u-boot-spl-ddr.map
ImagePos    Offset      Size  Name
00000000  00000000  00036f5c  main-section
00000000   00000000  0001ef5c  u-boot-spl
00000000    00000000  0001dc40  u-boot-spl-nodtb
0001dc40    0001dc40  0000131c  u-boot-spl-dtb
0001ef5c   0001ef5c  00008000  1d-imem
00026f5c   00026f5c  00004000  1d_dmem
0002af5c   0002af5c  00008000  2d_imem
00032f5c   00032f5c  00004000  2d_dmem

Tim,

Can you try deleting the &binman node from your mx8mm-venice-u-boot.dtsi file and see if the stock binman node in 8mm-u-boot-dtsi works for you? The new default one does for me on my Beacon 8mm board.

Quoted 6 lines

Tom,

I will submit another patch for the imx8mn_beacon boards to fix the build error on the 2g version. I forgot to update the second defconfig file when
I pushed the patch for the first one. I have it working, but I need to
clean it up.

OK, thanks! I just confirmed that with the above series it's down to:
imx8mn_beacon_2g imx8mm_venice kontron-sl-mx8mm phycore-imx8mm as the fails to build in CI boards:
https://source.denx.de/u-boot/u-boot/-/jobs/372239 as part of
https://source.denx.de/u-boot/u-boot/-/pipelines/10497

I feel at this point that worst-case I can come up with something to make venice and kontron link still given what you figured out above and shouldn't be any worse off than they are today. And that can go in v2022.01.

Signature · 1 line
Tom
#9 lore raw
Quoted 336 lines · Tom Rini
Quoted 7 lines · Adam Ford
Quoted 5 lines · Tom Rini
Quoted 4 lines · Adam Ford
Quoted 2 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <

[email protected]>

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines · wrote:
Quoted 8 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of

Tom

Quoted 5 lines
Quoted 5 lines

Rini

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit

generator

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

script"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

This reverts commit

d9a6f0eed66a39206b13513ec914f14084c3bb73.

Quoted 5 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

For right now, it's too close to the release to merge the

series

Quoted 2 lines
Quoted 2 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images,

and

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that

convert

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back

this

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator

script

Quoted 5 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with

mkimage

Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be complemented with

the

Quoted 9 lines
Quoted 9 lines

fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg):

/images/fdt@1

Quoted 3 lines
Quoted 3 lines

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg):

/images/fdt@2

Quoted 3 lines
Quoted 3 lines

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg):

/images/fdt@3

Quoted 3 lines
Quoted 3 lines

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg):

/images/fdt@4

Quoted 3 lines
Quoted 3 lines

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg):

/images/fdt@5

Quoted 3 lines
Quoted 3 lines

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg):

/images/atf@1

Quoted 10 lines
Quoted 10 lines

:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg

Quoted 8 lines
Quoted 8 lines

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg

property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image support

with

Quoted 10 lines
Quoted 10 lines

exit

Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I

stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to binman

for

Quoted 9 lines
Quoted 9 lines

this

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html

but I

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such

file or

Quoted 8 lines
Quoted 8 lines
Quoted 2 lines
Quoted 2 lines

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin'

failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch

resolved

Quoted 2 lines
Quoted 2 lines

the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also don't

know

Quoted 4 lines
Quoted 4 lines

what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did

not

Quoted 2 lines
Quoted 2 lines

get

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

caught by CI apparently because CI does a 'make all' which did

not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

include the FIT images (that was accomplished with the

'flash.bin'

Quoted 6 lines
Quoted 6 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman

conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1] and as

far

Quoted 5 lines
Quoted 5 lines

as I

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able

to

Quoted 4 lines
Quoted 4 lines

build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to

binman

Quoted 6 lines
Quoted 6 lines

[3]

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR =

"arch/arm/mach-imx/mkimage_fit_atf.sh":

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Tim
[1]

https://patchwork.ozlabs.org/project/uboot/list/?series=265765

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[2]

https://patchwork.ozlabs.org/project/uboot/list/?series=268380

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[3]

https://patchwork.ozlabs.org/project/uboot/list/?series=261640

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[4]

https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 10 lines
Quoted 10 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 2 lines
Quoted 2 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin

target

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were

not

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines

previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

and with this, we get some boards building and complaining as

expected:

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid
Quoted 5 lines
Quoted 5 lines
Quoted 3 lines
Quoted 3 lines

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR.

Please

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

migrate
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation

of

Quoted 4 lines
Quoted 4 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs

and

Quoted 14 lines
Quoted 14 lines

is

Quoted 13 lines

Review

Quoted 11 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just

Quoted 2 lines
Quoted 2 lines

grab

Quoted 1 line

the binman updates and apply it to master and fix it from that

starting

Quoted 5 lines
Quoted 5 lines
Quoted 1 line

point?

So, yes, grabbing the series from the above link and applying it on top of master, and making builds work with fake binaries would be very helpful. As best I can tell there's still something missing with

making

Quoted 1 line
Quoted 1 line

fake blobs link and not fail. For example, imx8mm_venice is part of

the

Quoted 3 lines
Quoted 3 lines

series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is
non-functional: lpddr4_pmu_train_1d_imem.bin

lpddr4_pmu_train_1d_dmem.bin

Quoted 5 lines
Quoted 5 lines

lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional:
atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps

with

Quoted 7 lines
Quoted 3 lines

previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman

node

Quoted 27 lines

from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

$ cat imx-boot.map
ImagePos    Offset      Size  Name
00000000  00000000  0011d778  main-section
00000000   00000000  00039000  spl
00057c00   00057c00  000c5b78  uboot
$ cat spl.map
ImagePos    Offset      Size  Name
00000000  00000000  00039000  main-section
00000000   00000000  00039000  mkimage
$ cat u-boot-spl-ddr.map
ImagePos    Offset      Size  Name
00000000  00000000  00036f5c  main-section
00000000   00000000  0001ef5c  u-boot-spl
00000000    00000000  0001dc40  u-boot-spl-nodtb
0001dc40    0001dc40  0000131c  u-boot-spl-dtb
0001ef5c   0001ef5c  00008000  1d-imem
00026f5c   00026f5c  00004000  1d_dmem
0002af5c   0002af5c  00008000  2d_imem
00032f5c   00032f5c  00004000  2d_dmem

Tim,

Quoted 8 lines

Can you try deleting the &binman node from your mx8mm-venice-u-boot.dtsi file and see if the stock binman node in 8mm-u-boot-dtsi works for you? The new default one does for me on my Beacon 8mm board.

Tom,

I will submit another patch for the imx8mn_beacon boards to fix the build error on the 2g version. I forgot to update the second defconfig file

Quoted 8 lines

when

Quoted 2 lines

I pushed the patch for the first one. I have it working, but I need to
clean it up.

OK, thanks! I just confirmed that with the above series it's down to:
imx8mn_beacon_2g imx8mm_venice kontron-sl-mx8mm phycore-imx8mm as the fails to build in CI boards:

I just submitted two patches against this WIP branch to remove the duplicate binman node from their respective u-boot.dtsi files as well. The kontron and phycore boards had the same issue as the venice board.

They were build-only tested.

Quoted 11 lines

https://source.denx.de/u-boot/u-boot/-/jobs/372239 as part of
https://source.denx.de/u-boot/u-boot/-/pipelines/10497

I feel at this point that worst-case I can come up with something to make venice and kontron link still given what you figured out above and shouldn't be any worse off than they are today. And that can go in v2022.01.

--
Tom

#10 lore raw
Quoted 294 lines · Adam Ford
Quoted 290 lines · Tom Rini
Quoted 287 lines · Adam Ford
Quoted 229 lines · Tom Rini
Quoted 6 lines · Adam Ford
Quoted 4 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 1 line · wrote:
Quoted 1 line
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom

Rini

Quoted 9 lines
Quoted 9 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the

series

Quoted 4 lines
Quoted 4 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

Quoted 8 lines
Quoted 8 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with

mkimage

Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be complemented with the

fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5

:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1

:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg

property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg

property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image support with

exit

Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I

stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to binman for

this

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html

but I

Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines

don't think that fully solves anything 'at this point' either.

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin'

failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved

the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also don't know

what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not

get

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman

conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1] and as far

as I

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to

build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon to binman

[3]

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
Quoted 6 lines

and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 5 lines
Quoted 5 lines
Quoted 3 lines

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 5 lines
Quoted 5 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 3 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin

target

Quoted 9 lines
Quoted 9 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 10 lines
Quoted 10 lines

and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid
Quoted 3 lines
Quoted 3 lines

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 5 lines

Review

Quoted 5 lines

+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please migrate

(quoted diff) +3 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and

is

Quoted 5 lines
Quoted 5 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':

Quoted 8 lines

Review

Quoted 6 lines

spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just

grab

Quoted 2 lines

the binman updates and apply it to master and fix it from that starting point?

Quoted 15 lines

So, yes, grabbing the series from the above link and applying it on top of master, and making builds work with fake binaries would be very helpful. As best I can tell there's still something missing with making fake blobs link and not fail. For example, imx8mm_venice is part of the series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is
non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional:
atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

$ cat imx-boot.map
ImagePos    Offset      Size  Name
00000000  00000000  0011d778  main-section
00000000   00000000  00039000  spl
00057c00   00057c00  000c5b78  uboot
$ cat spl.map
ImagePos    Offset      Size  Name
00000000  00000000  00039000  main-section
00000000   00000000  00039000  mkimage
$ cat u-boot-spl-ddr.map
ImagePos    Offset      Size  Name
00000000  00000000  00036f5c  main-section
00000000   00000000  0001ef5c  u-boot-spl
00000000    00000000  0001dc40  u-boot-spl-nodtb
0001dc40    0001dc40  0000131c  u-boot-spl-dtb
0001ef5c   0001ef5c  00008000  1d-imem
00026f5c   00026f5c  00004000  1d_dmem
0002af5c   0002af5c  00008000  2d_imem
00032f5c   00032f5c  00004000  2d_dmem

Tim,

Can you try deleting the &binman node from your mx8mm-venice-u-boot.dtsi file and see if the stock binman node in 8mm-u-boot-dtsi works for you? The new default one does for me on my Beacon 8mm board.

Tom,

I will submit another patch for the imx8mn_beacon boards to fix the build error on the 2g version. I forgot to update the second defconfig file when
I pushed the patch for the first one. I have it working, but I need to
clean it up.

Quoted 18 lines
Quoted 3 lines

OK, thanks! I just confirmed that with the above series it's down to:
imx8mn_beacon_2g imx8mm_venice kontron-sl-mx8mm phycore-imx8mm as the fails to build in CI boards:

I just submitted two patches against this WIP branch to remove the duplicate binman node from their respective u-boot.dtsi files as well. The kontron and phycore boards had the same issue as the venice board.

They were build-only tested.

Quoted 9 lines

https://source.denx.de/u-boot/u-boot/-/jobs/372239 as part of
https://source.denx.de/u-boot/u-boot/-/pipelines/10497

I feel at this point that worst-case I can come up with something to make venice and kontron link still given what you figured out above and shouldn't be any worse off than they are today. And that can go in v2022.01.

Tom,

I just resent my '[v3] imx8mm_venice: switch to use binman to pack images' as well (had to re-send as I forgot to include u-boot list earlier).

Tim

#11 lore raw
In reply to Adam Ford
Quoted 203 lines · Adam Ford
Quoted 188 lines · Tom Rini
Quoted 177 lines · Adam Ford
Quoted 175 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 10 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful. As best I can tell there's still something missing with making
fake blobs link and not fail. For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

Quoted 2 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Tim

Quoted 38 lines
$ cat imx-boot.map
ImagePos    Offset      Size  Name
00000000  00000000  0011d778  main-section
00000000   00000000  00039000  spl
00057c00   00057c00  000c5b78  uboot
$ cat spl.map
ImagePos    Offset      Size  Name
00000000  00000000  00039000  main-section
00000000   00000000  00039000  mkimage
$ cat u-boot-spl-ddr.map
ImagePos    Offset      Size  Name
00000000  00000000  00036f5c  main-section
00000000   00000000  0001ef5c  u-boot-spl
00000000    00000000  0001dc40  u-boot-spl-nodtb
0001dc40    0001dc40  0000131c  u-boot-spl-dtb
0001ef5c   0001ef5c  00008000  1d-imem
00026f5c   00026f5c  00004000  1d_dmem
0002af5c   0002af5c  00008000  2d_imem
00032f5c   00032f5c  00004000  2d_dmem

Tim,

Can you try deleting the &binman node from your mx8mm-venice-u-boot.dtsi file and see if the stock binman node in 8mm-u-boot-dtsi works for you? The new default one does for me on my Beacon 8mm board.

Tom,

I will submit another patch for the imx8mn_beacon boards to fix the build error on the 2g version. I forgot to update the second defconfig file when I pushed the patch for the first one. I have it working, but I need to clean it up.

adam

Quoted 5 lines

And that last error is the real problem.

--
Tom

author
#12 lore raw
Quoted 209 lines · Tim Harvey
Quoted 203 lines · Adam Ford
Quoted 188 lines · Tom Rini
Quoted 177 lines · Adam Ford
Quoted 175 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 10 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful. As best I can tell there's still something missing with making
fake blobs link and not fail. For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

Quoted 2 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too? The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Signature · 1 line
Tom
#13 lore raw
Quoted 219 lines · Tom Rini
Quoted 209 lines · Tim Harvey
Quoted 203 lines · Adam Ford
Quoted 188 lines · Tom Rini
Quoted 177 lines · Adam Ford
Quoted 175 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 10 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful. As best I can tell there's still something missing with making
fake blobs link and not fail. For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

Quoted 2 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Quoted 6 lines

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too? The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works. I will submit a v3 for you that does this.

Best regards,

Tim

author
#14 lore raw
Quoted 237 lines · Tim Harvey
Quoted 219 lines · Tom Rini
Quoted 209 lines · Tim Harvey
Quoted 203 lines · Adam Ford
Quoted 188 lines · Tom Rini
Quoted 177 lines · Adam Ford
Quoted 175 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g
(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 10 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful. As best I can tell there's still something missing with making
fake blobs link and not fail. For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext
Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

Quoted 2 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Quoted 6 lines

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too? The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Quoted 6 lines

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works. I will submit a v3 for you that does this.

Great, thanks!

Signature · 1 line
Tom
#15 lore raw
Quoted 331 lines · Tom Rini
Quoted 4 lines · Tim Harvey
Quoted 3 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Fri, Jan 7, 2022 at 12:08 PM Adam Ford <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

On Fri, Jan 7, 2022 at 12:25 PM Tom Rini <[email protected]>

Quoted 3 lines · wrote:
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines · Adam Ford
Quoted 1 line

On Fri, Jan 7, 2022 at 11:38 AM Tom Rini <[email protected]>

Quoted 3 lines · wrote:
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines · Tim Harvey
Quoted 1 line

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <

[email protected]>

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines · wrote:
Quoted 8 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On

Behalf Of Tom Rini

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit

generator

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

script"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

This reverts commit

d9a6f0eed66a39206b13513ec914f14084c3bb73.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

For right now, it's too close to the release to merge

the series

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final

images, and

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that

convert

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring

back this

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT

generator script

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible

with mkimage

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be complemented

with the fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg):

/images/fdt@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg):

/images/fdt@2:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg):

/images/fdt@3:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg):

/images/fdt@4:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg):

/images/fdt@5:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg):

/images/atf@1:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no

reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but no

reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no

reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no

reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no

reg property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image

support with exit

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when

(again, I stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to

binman for this

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

release). There was a patch that resolved this from

Oliver at

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

don't think that fully solves anything 'at this point'

either.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Even with that applied to current master I then end up

with:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No

such file or

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target

'flash.bin' failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch

resolved the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also

don't know what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures

did not get

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

caught by CI apparently because CI does a 'make all'

which did not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

include the FIT images (that was accomplished with the

'flash.bin'

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman

conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1] and

as far as I

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

know didn't get merged simply because of the CI issue. It

still

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was

able to build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves imx8mm_beacon

to binman [3]

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

and imx8mn_beacon to binman [4] but they no longer apply

due to

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR =

"arch/arm/mach-imx/mkimage_fit_atf.sh":

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Tim
[1]

https://patchwork.ozlabs.org/project/uboot/list/?series=265765

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[2]

https://patchwork.ozlabs.org/project/uboot/list/?series=268380

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[3]

https://patchwork.ozlabs.org/project/uboot/list/?series=261640

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[4]

https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 2 lines

Is it a show-stopper that it
does not build for boards using binman for release? From

what you

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

mentioned in another thread it was never building the

flash.bin target

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these boards

were not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

previously building the final image as part of "all" is a

problem.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

So, here's what I'm at right now. I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

and with this, we get some boards building and complaining as

expected:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs

and is

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs

and is

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR.

Please

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

migrate
+(imx8mn_beacon_2g) to binman instead, to avoid the

proliferation of

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing external

blobs and is

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Review

Quoted 6 lines

spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine. I can work on this one today. Do

I just grab

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

the binman updates and apply it to master and fix it from that

starting

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

point?

So, yes, grabbing the series from the above link and applying it

on top

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

of master, and making builds work with fake binaries would be very helpful. As best I can tell there's still something missing with

making

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

fake blobs link and not fail. For example, imx8mm_venice is part

of the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1

...':

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Image 'main-section:u-boot-spl-ddr' has faked external blobs and

is non-functional: lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Image 'main-section' is missing external blobs and is

non-functional: atf_blob blob-ext

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0)

overlaps with previous entry '/binman/imx-boot/uboot' ending at 0x1aee08 (1764872)

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

From what I can tell looking at the imx8mm_venice board is that

the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space. Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 1 line

Adam,

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not

sure

Quoted 1 line
Quoted 1 line
Quoted 1 line

if he added them yet in a way that was compatible with multiple

fdt's)

Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Right, so this is with

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 3 lines
Quoted 3 lines

applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too? The tree is also at

https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Quoted 12 lines
Quoted 1 line

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Quoted 9 lines
Quoted 6 lines

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works. I will submit a v3 for you that does this.

Great, thanks!

I just sent a patch for the beacon_imx8mn_2g board against your WIP branch with the rest of the binman stuff already applied. Do you need me to rebase that patch on master and squash it with the other nano? I wasn't sure if the WIP would get merged as-is with updates.

adam

Quoted 4 lines

--
Tom

#16 lore raw
In reply to Tim Harvey

Hi Tim and Tom

Quoted 247 lines · Tim Harvey
Quoted 229 lines · Tom Rini
Quoted 219 lines · Tim Harvey
Quoted 210 lines · Adam Ford
Quoted 192 lines · Tom Rini
Quoted 181 lines · Adam Ford
Quoted 179 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead.  So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
  MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
  MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper.  That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now.  I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
   aarch64:  w+   imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:
   aarch64:  +   imx8mn_beacon_2g

(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine.  I can work on this one today.  Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 13 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful.  As best I can tell there's still something missing with making
fake blobs link and not fail.  For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional:
lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin
Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx- boot/uboot' ending at 0x1aee08 (1764872)

Quoted 5 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space.  Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Quoted 6 lines

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too?  The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Quoted 5 lines

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works.

I also quickly double checked everything and indeed somehow the removal of the binman stuff from arch/arm/dts/imx8mm-venice-u-boot.dtsi which I had in my 4th patch "[PATCH v3 4/7] arm64: dts: imx8mm: use common binman configuration" got lost. Otherwise it all looks good. Thanks!

Quoted 5 lines

I will submit a v3 for you that does this.

Best regards,

Tim

Cheers

Marcel

author
#17 lore raw
Quoted 253 lines · Marcel Ziswiler

Hi Tim and Tom

Quoted 250 lines · Tim Harvey
Quoted 229 lines · Tom Rini
Quoted 219 lines · Tim Harvey
Quoted 210 lines · Adam Ford
Quoted 192 lines · Tom Rini
Quoted 181 lines · Adam Ford
Quoted 179 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead.  So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
  MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
  MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper.  That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now.  I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
   aarch64:  w+   imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:
   aarch64:  +   imx8mn_beacon_2g

(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine.  I can work on this one today.  Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 13 lines

So, yes, grabbing the series from the above link and applying it on top
of master, and making builds work with fake binaries would be very
helpful.  As best I can tell there's still something missing with making
fake blobs link and not fail.  For example, imx8mm_venice is part of the
series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional:
lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin
Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx- boot/uboot' ending at 0x1aee08 (1764872)

Quoted 5 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are two copies trying to occupy the same space.  Deleting the &binman node from the venice-u-boot.dtsi file appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Quoted 6 lines

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too?  The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

Quoted 6 lines
Quoted 2 lines

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works.

I also quickly double checked everything and indeed somehow the removal of the binman stuff from arch/arm/dts/imx8mm-venice-u-boot.dtsi which I had in my 4th patch "[PATCH v3 4/7] arm64: dts: imx8mm: use common binman configuration" got lost. Otherwise it all looks good. Thanks!

I must have messed it up, sorry. I'm doing one more CI run right now of Adam's versions of the patches (just to be extra safe) and then I'll merge the branch. Thanks all!

Signature · 1 line
Tom
#18 lore raw
Quoted 356 lines · Tom Rini
Quoted 7 lines · Marcel Ziswiler

Hi Tim and Tom

Quoted 4 lines · Tim Harvey
Quoted 3 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Fri, Jan 7, 2022 at 12:08 PM Adam Ford <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

On Fri, Jan 7, 2022 at 12:25 PM Tom Rini <[email protected]>

Quoted 3 lines · wrote:
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines · Adam Ford
Quoted 1 line

On Fri, Jan 7, 2022 at 11:38 AM Tom Rini <[email protected]>

Quoted 3 lines · wrote:
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines · Tim Harvey
Quoted 1 line

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <

[email protected]>

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines · wrote:
Quoted 8 lines
Quoted 5 lines · ZHIZHIKIN Andrey

Hello Tom,

Quoted 2 lines

-----Original Message-----
From: U-Boot <[email protected]> On

Behalf Of Tom Rini

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit

generator

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

script"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

This reverts commit

d9a6f0eed66a39206b13513ec914f14084c3bb73.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

For right now, it's too close to the release to

merge the series

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final

images, and

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series

that convert

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead. So, bring

back this

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT

generator script

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not

compatible with mkimage

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

due

Quoted 1 line
Quoted 1 line
Quoted 1 line

to CVE-2021-27138. This revert should be

complemented with the fix to

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

remove those '@' symbols as well.

Correct, the revert is not enough anymore:
MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg):

/images/uboot@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg):

/images/fdt@1:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg):

/images/fdt@2:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg):

/images/fdt@3:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg):

/images/fdt@4:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg):

/images/fdt@5:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg):

/images/atf@1:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but

no reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):
/configurations/config@2: node has a unit name, but

no reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but

no reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but

no reg property

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but

no reg property

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

./tools/mkimage: verify_header failed for FIT Image

support with exit

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

code 1

Quoted 6 lines
Quoted 6 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when

(again, I stopped

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

worrying about it because I thought we were moving to

binman for this

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

release). There was a patch that resolved this from

Oliver at

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

don't think that fully solves anything 'at this point'

either.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Even with that applied to current master I then end up

with:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No

such file or

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

directory

Quoted 1 line
Quoted 1 line

arch/arm/mach-imx/Makefile:167: recipe for target

'flash.bin' failed

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin' Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that

patch resolved the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

building issue when using the FIT generator but I also

don't know what

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

else has changed that now causes that to not work.

As Tom pointed out in another thread these build

failures did not get

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

caught by CI apparently because CI does a 'make all'

which did not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

include the FIT images (that was accomplished with the

'flash.bin'

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

target prior to binman conversion).

Is it too late to apply the CI fix and the pending

binman conversions?

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I know that my series has been reviewed by Marcel [1]

and as far as I

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

know didn't get merged simply because of the CI issue.

It still

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which

converts

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and

was able to build

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

flash.bin images for them

I tried to merge Adam's series that moves

imx8mm_beacon to binman [3]

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

and imx8mn_beacon to binman [4] but they no longer

apply due to

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

defconfig/Kconfig changes

That still leaves the following unbuildable with
CONFIG_SPL_FIT_GENERATOR =

"arch/arm/mach-imx/mkimage_fit_atf.sh":

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

imx/mkimage_fit_atf.sh"

Quoted 3 lines
Quoted 3 lines

Tim
[1]

https://patchwork.ozlabs.org/project/uboot/list/?series=265765

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[2]

https://patchwork.ozlabs.org/project/uboot/list/?series=268380

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[3]

https://patchwork.ozlabs.org/project/uboot/list/?series=261640

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

[4]

https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Tom,

I'm not familiar with the U-boot CI tool.

It's at

https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

(and in-tree under doc/develop/ci_testing.rst).

Quoted 2 lines

Is it a show-stopper that it
does not build for boards using binman for release? From

what you

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

mentioned in another thread it was never building the

flash.bin target

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper. That all of these

boards were not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

previously building the final image as part of "all" is a

problem.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

So, here's what I'm at right now. I've grabbed Heiko's

patch.

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

and with this, we get some boards building and complaining

as expected:

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

aarch64: w+ imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external

blobs and is

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_evk) Image 'main-section' is missing external

blobs and is

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:

   aarch64:  +   imx8mn_beacon_2g

+(imx8mn_beacon_2g) ===================== WARNING

======================

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) This board uses

CONFIG_SPL_FIT_GENERATOR. Please

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

migrate
+(imx8mn_beacon_2g) to binman instead, to avoid the

proliferation of

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g)

====================================================

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) Image 'main-section' is missing

external blobs and is

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl .cfgout -T imx8mimage -e 0x912000

./mkimage-out.spl.mkimage':

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Review

Quoted 4 lines

spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

(quoted diff) +2 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all]

Error 1

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make]

Error 2

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

The beacon boards are mine. I can work on this one today.

Do I just grab

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

the binman updates and apply it to master and fix it from

that starting

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

point?

So, yes, grabbing the series from the above link and applying

it on top

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

of master, and making builds work with fake binaries would be

very

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

helpful. As best I can tell there's still something missing

with making

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

fake blobs link and not fail. For example, imx8mm_venice is

part of the

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

series above but still fails with 'make

BINMAN_FAKE_EXT_BLOBS=1 ...':

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Image 'main-section:u-boot-spl-ddr' has faked external blobs

and is non-functional:

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin

lpddr4_pmu_train_2d_imem.bin

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

lpddr4_pmu_train_2d_dmem.bin
Image 'main-section' is missing external blobs and is

non-functional: atf_blob blob-ext

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0)

overlaps with previous entry '/binman/imx-

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 2 lines

boot/uboot' ending at 0x1aee08 (1764872)

From what I can tell looking at the imx8mm_venice board is that

the imx8mm-u-boot.dtsi has the binman

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

nodes to build the image, but

arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

two copies trying to occupy the same space. Deleting the

&binman node from the venice-u-boot.dtsi file

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

appears to make the problem go away, and binman is still making

the image.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 1 line

Adam,

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not

sure

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

if he added them yet in a way that was compatible with multiple

fdt's)

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Right, so this is with

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/

Quoted 1 line
Quoted 1 line
Quoted 1 line

applied and I did have to manually apply Marcel's 4/7 and 7/7, so

maybe

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I mismerged something there too? The tree is also at

https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Quoted 12 lines
Quoted 12 lines
Quoted 1 line

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Quoted 7 lines
Quoted 5 lines

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works.

I also quickly double checked everything and indeed somehow the removal

of the binman stuff from

Quoted 9 lines
Quoted 1 line

arch/arm/dts/imx8mm-venice-u-boot.dtsi which I had in my 4th patch

"[PATCH v3 4/7] arm64: dts: imx8mm: use

Quoted 1 line

common binman configuration" got lost. Otherwise it all looks good.

Thanks!

I must have messed it up, sorry. I'm doing one more CI run right now of Adam's versions of the patches (just to be extra safe) and then I'll merge the branch. Thanks all!

Thanks for your work and helping us get this merged while the imx maintainer is out.

adam

Quoted 4 lines

--
Tom

#19 lore raw
In reply to Tom Rini

Hi Tom

Quoted 260 lines · Tom Rini
Quoted 259 lines · Marcel Ziswiler

Hi Tim and Tom

Quoted 256 lines · Tim Harvey
Quoted 235 lines · Tom Rini
Quoted 225 lines · Tim Harvey
Quoted 213 lines · Adam Ford
Quoted 194 lines · Tom Rini
Quoted 183 lines · Adam Ford
Quoted 181 lines · Tom Rini
Quoted 1 line · Tim Harvey

On Thu, Jan 6, 2022 at 12:27 PM Tim Harvey <[email protected]>

Quoted 4 lines · wrote:
Quoted 4 lines
Quoted 1 line · ZHIZHIKIN Andrey
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Hello Tom,

Quoted 6 lines

-----Original Message-----
From: U-Boot <[email protected]> On Behalf Of Tom Rini
Sent: Thursday, January 6, 2022 7:52 PM
To: [email protected]
Cc: Tim Harvey <[email protected]>
Subject: [PATCH] Revert "tree: imx: remove old fit generator

script"

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

This reverts commit d9a6f0eed66a39206b13513ec914f14084c3bb73.

For right now, it's too close to the release to merge the series

that

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

allows for binman to be used to generate the final images, and

also not

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

break CI, and then also merge all of the series that convert

currently

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

broken platforms to use binman instead.  So, bring back this

script now

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line

and remove it again for real after the release.

Please note that this might not work, as the FIT generator script

would

Quoted 1 line
Quoted 1 line
Quoted 1 line

generate ITS with '@' symbols which are not compatible with mkimage

due

Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

to CVE-2021-27138. This revert should be complemented with the fix to remove those '@' symbols as well.

Correct, the revert is not enough anymore:
  MKIMAGE u-boot.itb
u-boot.its:7.11-15.5: Warning (unit_address_vs_reg): /images/uboot@1:
node has a unit name, but no reg property u-boot.its:16.9-21.5: Warning (unit_address_vs_reg): /images/fdt@1:
node has a unit name, but no reg property u-boot.its:22.9-27.5: Warning (unit_address_vs_reg): /images/fdt@2:
node has a unit name, but no reg property u-boot.its:28.9-33.5: Warning (unit_address_vs_reg): /images/fdt@3:
node has a unit name, but no reg property u-boot.its:34.9-39.5: Warning (unit_address_vs_reg): /images/fdt@4:
node has a unit name, but no reg property u-boot.its:40.9-45.5: Warning (unit_address_vs_reg): /images/fdt@5:
node has a unit name, but no reg property u-boot.its:46.9-55.5: Warning (unit_address_vs_reg): /images/atf@1:
node has a unit name, but no reg property u-boot.its:60.12-65.5: Warning (unit_address_vs_reg):
/configurations/config@1: node has a unit name, but no reg property u-boot.its:66.12-71.5: Warning (unit_address_vs_reg):

Quoted 8 lines
Quoted 8 lines

/configurations/config@2: node has a unit name, but no reg property u-boot.its:72.12-77.5: Warning (unit_address_vs_reg):
/configurations/config@3: node has a unit name, but no reg property u-boot.its:78.12-83.5: Warning (unit_address_vs_reg):
/configurations/config@4: node has a unit name, but no reg property u-boot.its:84.12-89.5: Warning (unit_address_vs_reg):
/configurations/config@5: node has a unit name, but no reg property ./tools/mkimage: verify_header failed for FIT Image support with exit

code 1

Quoted 11 lines
Quoted 11 lines

Makefile:1433: recipe for target 'u-boot.itb' failed
make: *** [u-boot.itb] Error 1
make: *** Deleting file 'u-boot.itb'
make: *** Waiting for unfinished jobs....

I don't know what had changed to cause this or when (again, I stopped worrying about it because I thought we were moving to binman for this release). There was a patch that resolved this from Oliver at
https://lists.denx.de/pipermail/u-boot/2021-August/457997.html but I don't think that fully solves anything 'at this point' either.

Quoted 3 lines
Quoted 3 lines

Even with that applied to current master I then end up with:
  MKIMAGE flash.bin
./tools/mkimage: Can't open spl/u-boot-spl-ddr.bin: No such file or

directory

Quoted 23 lines
Quoted 23 lines

arch/arm/mach-imx/Makefile:167: recipe for target 'flash.bin' failed make[1]: *** [flash.bin] Error 1
make[1]: *** Deleting file 'flash.bin'
Makefile:1526: recipe for target 'flash.bin' failed

At some point over the past couple of months that patch resolved the building issue when using the FIT generator but I also don't know what else has changed that now causes that to not work.

As Tom pointed out in another thread these build failures did not get caught by CI apparently because CI does a 'make all' which did not include the FIT images (that was accomplished with the 'flash.bin' target prior to binman conversion).

Is it too late to apply the CI fix and the pending binman conversions?

I know that my series has been reviewed by Marcel [1] and as far as I know didn't get merged simply because of the CI issue. It still applies and produces a valid flash.bin image.
I was also able to merge Peng's series [2] which converts imx8mq_evk/imx8mq_phanbell/pico-imx8mq to binman and was able to build flash.bin images for them

Quoted 7 lines
Quoted 7 lines

I tried to merge Adam's series that moves imx8mm_beacon to binman [3] and imx8mn_beacon to binman [4] but they no longer apply due to defconfig/Kconfig changes

That still leaves the following unbuildable with CONFIG_SPL_FIT_GENERATOR = "arch/arm/mach-imx/mkimage_fit_atf.sh":

configs/cgtqmx8_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach-imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-ctouch2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm-icore-mx8mm-edimm2.2_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mm_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_2g_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8mn_beacon_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 1 line
Quoted 1 line

configs/imx8qm_rom7720_a1_4G_defconfig:CONFIG_SPL_FIT_GENERATOR="arch/arm/mach- imx/mkimage_fit_atf.sh"

Quoted 11 lines
Quoted 7 lines

Tim
[1] https://patchwork.ozlabs.org/project/uboot/list/?series=265765
[2] https://patchwork.ozlabs.org/project/uboot/list/?series=268380
[3] https://patchwork.ozlabs.org/project/uboot/list/?series=261640
[4] https://patchwork.ozlabs.org/project/uboot/list/?series=261822

Tom,

I'm not familiar with the U-boot CI tool.

It's at https://u-boot.readthedocs.io/en/latest/develop/ci_testing.html
(and in-tree under doc/develop/ci_testing.rst).

Quoted 4 lines

Is it a show-stopper that it
does not build for boards using binman for release? From what you mentioned in another thread it was never building the flash.bin target for the boards using the FIT generator anyway.

Yes, breaking CI is a ship-stopper.  That all of these boards were not previously building the final image as part of "all" is a problem.

So, here's what I'm at right now.  I've grabbed Heiko's patch. Everything is currently visible at

https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
and with this, we get some boards building and complaining as expected:
   aarch64:  w+   imx8mn_evk
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext@1
blob-ext@2 blob-ext@3 blob-ext@4
+(imx8mn_evk) Image 'main-section' is missing external blobs and is
non-functional: blob-ext

(quoted diff) +2 -0
+(imx8mn_evk)
+(imx8mn_evk) Some images are invalid

But others are:
   aarch64:  +   imx8mn_beacon_2g

(quoted diff) +2 -0
+(imx8mn_beacon_2g) ===================== WARNING ======================
+(imx8mn_beacon_2g) This board uses CONFIG_SPL_FIT_GENERATOR. Please

migrate

(quoted diff) +4 -0
+(imx8mn_beacon_2g) to binman instead, to avoid the proliferation of
+(imx8mn_beacon_2g) arch-specific scripts with no tests.
+(imx8mn_beacon_2g) ====================================================
+(imx8mn_beacon_2g) Image 'main-section' is missing external blobs and is

non-functional: blob-
ext@1 blob-ext@2 blob-ext@3 blob-ext@4 +(imx8mn_beacon_2g) binman: Error 1 running 'mkimage -d ./mkimage.spl.mkimage -n spl/u-boot-spl
.cfgout -T imx8mimage -e 0x912000 ./mkimage-out.spl.mkimage':
spl/u-boot-spl-ddr.bin: Can't ope
n: No such file or directory

Quoted 8 lines

Review

Quoted 4 lines
(quoted diff) +3 -0
+(imx8mn_beacon_2g)
+(imx8mn_beacon_2g) make[1]: *** [Makefile:1088: all] Error 1
+(imx8mn_beacon_2g) make: *** [Makefile:177: sub-make] Error 2

The beacon boards are mine.  I can work on this one today.  Do I just grab the binman updates and apply it to master and fix it from that starting point?

Quoted 14 lines

So, yes, grabbing the series from the above link and applying it on top of master, and making builds work with fake binaries would be very helpful.  As best I can tell there's still something missing with making fake blobs link and not fail.  For example, imx8mm_venice is part of the series above but still fails with 'make BINMAN_FAKE_EXT_BLOBS=1 ...':
Image 'main-section:u-boot-spl-ddr' has faked external blobs and is non-functional:
lpddr4_pmu_train_1d_imem.bin lpddr4_pmu_train_1d_dmem.bin lpddr4_pmu_train_2d_imem.bin lpddr4_pmu_train_2d_dmem.bin
Image 'main-section' is missing external blobs and is non-functional: atf_blob blob-ext Wrote map file './imx-boot.map' to show errors
binman: Node '/binman/imx-boot/blob-ext@1': Offset 0x0 (0) overlaps with previous entry '/binman/imx-
boot/uboot' ending at 0x1aee08 (1764872)

Quoted 8 lines

From what I can tell looking at the imx8mm_venice board is that the imx8mm-u-boot.dtsi has the binman
nodes to build the image, but arch/arm/dts/imx8mm-venice-u-boot.dtsi duplicates that work, so there are
two copies trying to occupy the same space.  Deleting the &binman node from the venice-u-boot.dtsi file
appears to make the problem go away, and binman is still making the image.

Adam,

Quoted 7 lines

I don't quite follow. I don't see any binman nodes in arch/arm/dts/imx8mm-u-boot.dtsi.

I know that Marcel submitted a patch that added them (and I'm not sure if he added them yet in a way that was compatible with multiple fdt's) but this hasn't been merged yet so I don't see any duplicate binman nodes for venice.

Quoted 6 lines

Right, so this is with
https://patchwork.ozlabs.org/bundle/trini/2022-01-07-imx8-and-binman-updates/
applied and I did have to manually apply Marcel's 4/7 and 7/7, so maybe I mismerged something there too?  The tree is also at
https://github.com/trini/u-boot/tree/WIP/2022-01-07-imx8-and-buildman-updates

Tom,

Thank you for helping with this. During the merge window we had a lot of collisions because of things moving to defconfig that I think caused a lot of imx patches to have to be rebased. Then as most of us were submitting binman patches this CI issue popped up and as Stefano was unavailable I think it all got left hanging. In the middle of that Marcel's patch moved binman nodes to a common place so some rebasing is needed there as well. Fun times!

Yes, I couldn't get all from your patchwork bundle to apply - I didn't try as hard as you to manually merge :)

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

I pulled your tree and see Marcel's patches there now. With the binman node removed from arch/arm/dts/imx8mm-venice-u-boot.dtsi venice works.

I also quickly double checked everything and indeed somehow the removal of the binman stuff from arch/arm/dts/imx8mm-venice-u-boot.dtsi which I had in my 4th patch "[PATCH v3 4/7] arm64: dts: imx8mm: use common binman configuration" got lost. Otherwise it all looks good. Thanks!

I must have messed it up, sorry.

No, I guess you just applied Tim's and my patches in a different order and therefore this binman stuff removal for venice wasn't really required resp. not even there yet. And Tim's later patch already omitted it. So all looks indeed fine.

Quoted 3 lines

I'm doing one more CI run right now of
Adam's versions of the patches (just to be extra safe) and then I'll merge the branch.  Thanks all!

Thank you!

Cheers

Marcel