How to confirm image installed by MCU boot

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Wed, 02 Apr 2025 12:59:47 GMT

Hi,

yes, we got something to run, so that we are at least able to update firmware. Logging would be cool, but we decided (not recommended!) to use RTT for logging and that is very unreliable (see https://devzone.nordicsemi.com/f/nordic-q-a/116860/unstable-rtt-logging)

you can close this issue.

thanks and best regards

Torsten

RE: How to confirm image installed by MCU boot

Menon — Wed, 02 Apr 2025 12:53:58 GMT

Hello Torsten,

Sorry for the late response. Were you able to get this working, or are you still facing issues? Let me know the current status, or update me with the latest error if the problem is still ongoing.

Just to clarify: CONFIG_USE_NRF53_MULTI_IMAGE_WITHOUT_UPGRADE_ONLY enables non-upgrade-only multi-image updates, but it doesn’t resolve rollback support for the network core. That part won’t work unless CONFIG_BOOT_UPGRADE_ONLY=n.

[quote user="Torsten Robitzki"]mcuboot: Image in the secondary slot is not valid!` [/quote]

Try building with MCUboot verbose logging enabled, it will help identify exactly what is failing. This will provide more details on whether the issue is related to signature verification, decryption, or the image structure.

Kind Regards,

Abhijith

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Tue, 18 Feb 2025 15:55:19 GMT

Hi Menon ,

I tried to apply your solution. I inserted `CONFIG_USE_NRF53_MULTI_IMAGE_WITHOUT_UPGRADE_ONLY=y` into `prj.conf`. Trying to add that to `sysbuild.conf` resulted in an error. I'm not sure, what you mean by "project-level Kconfig file" (same applies to CONFIG_USE_NRF53_MULTI_IMAGE_WITHOUT_UPGRADE_ONLY).

Now, the bootloader fails with the log message `mcuboot: Image in the secondary slot is not valid!` with the first image slot (application).

I tried both, the signed and the signed and encrypted image. The image header and trailer seem to be ok (verfied with a debugger and I get the "magic=good" log message, that confirms the image trailer). I also confirmed, that the bootloader is reading the image correctly by attaching a logic analyser to the flash memory and looking at the traces.

Using imgtool to verify the signed image, works. Using imgtool to verify the signed and encrypted image fails! I don't know, if that is on purpose or not. The image header for the signed and encrypted image looks like this:

Printing content of signed image: zephyr.signed.encrypted.bin

#### Image header (offset: 0x0) ############################
magic: 0x96f3b83d
load_addr: 0x0
hdr_size: 0x200
protected_tlv_size: 0x0
img_size: 0xba1a0
flags: ENCRYPTED_AES128 (0x4)
version: 0.1.42+0
############################################################
#### Payload (offset: 0x200) ###############################
| |
| FW image (size: 0xba1a0 Bytes) |
| |
############################################################
#### TLV area (offset: 0xba3a0) ############################
magic: 0x6907
area size: 0x10c
---------------------------------------------
type: SHA256 (0x10)
len: 0x20
data: 0x07 0x46 0x19 0xbd 0x1f 0x8d 0xde 0xee
0x38 0xde 0x79 0x93 0x0f 0x94 0x8a 0x45
0x7b 0x62 0xd7 0xab 0x86 0x39 0x7b 0x6a
0x5c 0xa6 0xfb 0x7a 0x03 0xf0 0xcd 0x71
---------------------------------------------
type: KEYHASH (0x1)
len: 0x20
data: 0x3e 0xb0 0x2c 0xc0 0xb5 0xdc 0x6e 0x94
0x9d 0x26 0xb3 0xad 0x4d 0x10 0xa2 0x62
0x94 0xd9 0xf4 0x3e 0x5f 0x49 0xaf 0x89
0x11 0xb0 0x21 0x81 0xe1 0x9d 0x99 0x2d
---------------------------------------------
type: ECDSASIG (0x22)
len: 0x47
data: 0x30 0x45 0x02 0x21 0x00 0xf7 0x49 0x32
0x8c 0x61 0x2d 0xd0 0x40 0xd4 0x66 0x49
0x2d 0xad 0xc4 0x91 0xa0 0xb2 0xb9 0x84
0x2d 0x92 0xd2 0x30 0x36 0x25 0xcc 0xcf
0x3b 0xb4 0xa3 0xce 0xab 0x02 0x20 0x62
0xa6 0x7e 0xe9 0xfd 0x18 0xd7 0x03 0xce
0xe8 0x83 0x97 0x6c 0x4f 0x9b 0x03 0x75
0xea 0xa4 0x46 0x2c 0x9f 0x6b 0x27 0x7e
0x8d 0x11 0x20 0x84 0xec 0x67 0x43
---------------------------------------------
type: ENCEC256 (0x32)
len: 0x71
data: 0x04 0x86 0x1c 0xba 0x9a 0x6d 0x7e 0xc2
0x08 0x4c 0xe2 0xcb 0x75 0xac 0x74 0xb8
0xf5 0x89 0x11 0xdf 0xf9 0x5d 0xd6 0xc6
0x73 0x7e 0x34 0xc2 0xd8 0xb5 0x7a 0x82
0x39 0x4f 0xb9 0xd0 0x0b 0xd9 0x60 0x46
0x7c 0x43 0x40 0x4f 0x0b 0x8c 0x9f 0x46
0x15 0xa5 0x4a 0x22 0xf7 0x2e 0xa3 0x20
0xcf 0x6e 0x2a 0x6d 0x60 0x30 0x48 0x79
0x8f 0xc1 0xab 0x27 0x7b 0x25 0xf2 0xd1
0x7d 0x96 0x19 0xe1 0x59 0xaf 0x6d 0x49
0x77 0x3e 0x17 0x03 0x0c 0x0d 0xbc 0xd0
0x0e 0xfe 0xaa 0x53 0xef 0x1c 0x1e 0x51
0x34 0xce 0x6b 0x72 0xa2 0xad 0xd5 0x44
0x1d 0xf9 0xd0 0x77 0x51 0x9f 0xc2 0x7a
0xaf
############################################################
#### End of Image ##########################################
dumpinfo has run successfully

According to the debugger, the correct private key is stored in the bootloader (in bootutil_enc_key).

What can be the issue here? How to get even more details as to what causes the failure?

Thanks in advance and best regards

Torsten

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Tue, 21 Jan 2025 09:36:26 GMT

Hi Menon ,

no problem, we are also very busy at the moment. I've added this issue to our ticket system and it might take some time, before someone is able to work on this further.

thanks

Torsten

RE: How to confirm image installed by MCU boot

Menon — Fri, 10 Jan 2025 07:21:10 GMT

Hello Torsten,

Sorry for getting back to you so late. I took a long vacation and resumed work this week. Usually, some of my colleagues would take over in my absence, but I guess we were very understaffed during the Christmas period, which caused this case to remain pending for a long time.

Regarding your issue:

CONFIG_BOOT_SWAP_USING_MOVE=y: The use of Swap is not recommended as MCUBoot does not support image swapping (test swaps) for the network core due to the way RAM and PCD interact during reversion.
CONFIG_BOOT_UPGRADE_ONLY: The simultaneous update of multiple images does not support network core image reversion, so you need to disable application image reversion.

The application core can be reverted, but doing so breaks the network core upon reversal, as the reversion process overwrites the network core with the content currently in the RAM that PCD uses. To enable this, define the CONFIG_USE_NRF53_MULTI_IMAGE_WITHOUT_UPGRADE_ONLY Kconfig option in the project-level Kconfig file. Once defined, you can activate it by setting CONFIG_USE_NRF53_MULTI_IMAGE_WITHOUT_UPGRADE_ONLY.

Refer Simultaneous multi-image DFU with nRF5340

Kind Regards,

Abhijith

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Mon, 30 Dec 2024 13:49:15 GMT

Hi Menon ,

I hope you are doing well. Does you checkings had any results? Will that issue be fixed? Any timeline?

best regards,

Torsten

RE: How to confirm image installed by MCU boot

Menon — Tue, 12 Nov 2024 16:42:14 GMT

Hello,

I wanted to let you know that I am still working on this case and am checking internally regarding this issue.

I will keep you updated.

Kind regards,
Abhijith

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Thu, 07 Nov 2024 17:58:09 GMT

Yes, after the first reboot, the log messages from MCUBoot suggest, that it performs a test swap installation and after confirming the image, and a next reboot, the image is not switched back. If I skip the confirmation, the image get switched back after rebooting.

Removing the offensive line of code, was just a test. I really would hate to have clones of all the zephyr and nordic repositories for all this kind of hacks. When possible, I would love to see that change in one of the next releases, so that I could just upgrade to the next version or, of cause, having some other solution, that would make that code change unnecessary.

best regards

Torsten

RE: How to confirm image installed by MCU boot

Menon — Thu, 07 Nov 2024 17:29:37 GMT

Hello,

[quote user="Torsten Robitzki"]Maybe, you could check with some of the developers, if that text is really necessary?[/quote]

Glad to hear it's working! I can check internally to see if removing the #error directive in mcuboot_config.h is necessary.

But have you tested the MCUboot mechanism after removing the #error? I recommend checking if the device boots into the new image and doesn't revert to the old one (maybe try rebooting the device a few times).

Kind Regards,

Abhijith

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Tue, 05 Nov 2024 15:34:40 GMT

Hi,

sorry, it's me again. I figured out, that `CONFIG_MCUBOOT_BOOTUTIL_LIB` requires also, `CONFIG_IMG_MANAGER` to be set to `y`, otherwise, `zephyr/subsys/dfu/CMakeLists.txt` is not added as subdirectory in the CMake configuration step.

`CONFIG_IMG_MANAGER` has a dependency to `CONFIG_STREAM_FLASH`:

warning: IMG_MANAGER (defined at subsys/dfu/Kconfig:10) was assigned the value 'y' but got the value
'n'. Check these unsatisfied dependencies: STREAM_FLASH (=n). See
http://docs.zephyrproject.org/latest/kconfig.html#CONFIG_IMG_MANAGER and/or look up IMG_MANAGER in
the menuconfig/guiconfig interface. The Application Development Primer, Setting Configuration
Values, and Kconfig - Tips and Best Practices sections of the manual might be helpful too.

After also adding that configuration, I ended up again with a compilation error:

/../../zephyr/include/mcuboot_config/mcuboot_config.h:75:2: error: #error "CONFIG_BOOT_SWAP_USING_MOVE not set"
   75 | #error "CONFIG_BOOT_SWAP_USING_MOVE not set"
      |  ^~~~~

When adding `CONFIG_BOOT_SWAP_USING_MOVE=y` to the configuration, then KConfig complains about the undefined configuration symbol:

prj.conf:84: warning: attempt to assign the value 'y' to the undefined symbol BOOT_SWAP_USING_MOVE
Parsing /Users/todi/bloomlife/zephyr-workspace/zephyr/Kconfig
Loaded configuration '/.../zephyr-workspace/zephyr/boards/nordic/nrf5340dk/nrf5340dk_nrf5340_cpuapp_defconfig'
Merged configuration '/.../zephyr-workspace/lovelace-pod-firmware/prj.conf'
Merged configuration '/.../zephyr-workspace/lovelace-pod-firmware/build/lovelace-pod-firmware/zephyr/.config.sysbuild'

In the end, I really don't need `boot_write_img_confirmed_multi()` from `zephyr/subsys/dfu/boot/mcuboot.c`, as this function really does nothing else, but to call `boot_set_confirmed_multi()` from the MCUBoot subtree.

It would probably helpful, if the dependencies that are encoded in some parts of the CMake and C code, would also be reflected in the Kconfig configuration.

Any idea, what I could try next?

best regards

Torsten

BTW: Removing that `#error` preprocessor statement in `mcuboot_config.h` would solve that issue for me. Maybe, you could check with some of the developers, if that text is really necessary?

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Tue, 05 Nov 2024 14:07:59 GMT

Hi Abhijith,

looks like this is working. At least, this builds and I also expect it to work at runtime.

Thank you very much,

Torsten

RE: How to confirm image installed by MCU boot

Menon — Tue, 05 Nov 2024 13:42:57 GMT

Hello Torsten,

I was looking more into this, and I think the config CONFIG_MCUBOOT_BOOTUTIL_LIB cannot be directly enabled. Instead, it is selected by enabling CONFIG_MCUBOOT_IMG_MANAGER. Try enabling this configuration inside the prj.conf and see if it makes any changes.

Kind Regards,

Abhijith

RE: How to confirm image installed by MCU boot

Menon — Tue, 05 Nov 2024 10:00:34 GMT

Hello Torsten,

Sorry if my answer was not clear to you. I didn't ask to check your configuration files. I can already see that you are enabling that in your project (which I believe is a correct method), instead I was pointing to the build configuration, i.e. to verify if the configuration you enabled has been properly reflected after the build. Sometimes even if you enabled, this might not get reflected, causing errors. For example if you check the file build\mcuboot\zephyr-->.config you could see which ever configs get enabled after the build. For the application, check build\zephyr-->.config, see if SB_CONFIG_BOOT_SWAP_USING_MOVE=y is getting enabled in any of the build configs.

I haven't tested this on my end after the introduction of Sysbuild, but I am not sure why you felt the answer was from Chatgpt. I have only asked to verify if configs are getting properly enabled after the build (asked this since the error messages says so). If you are still confused please share the build folder and I can have a look.

Kind Regards,

Abhijith

RE: How to confirm image installed by MCU boot

Torsten Robitzki — Tue, 05 Nov 2024 07:52:02 GMT

Hello Abhijith,

to be honest, I do not understand your answer. I've showed, what I've tried so far and using the configuration above, results in an error message. The configuration in question is even cited and you agree with me, that it should work.

But: It doesn't work. I applied the very same change to an example given to my by one of your colleagues (https://devzone.nordicsemi.com/f/nordic-q-a/115570/using-mcuboot-with-nrf5340 peripheral_lbs_enc_fota.zip) and it results in the very same error message.

If you say, I should check the configuration files: What is the expected configuration?

To me, it looks like, I received an answer from ChatGPT. I apologize, if I'm wrong.

best regards

Torsten

RE: How to confirm image installed by MCU boot

Menon — Mon, 04 Nov 2024 19:32:54 GMT

Hello,

[quote user=""]When I try to instead add `SB_CONFIG_BOOT_SWAP_USING_MOVE=y` to sysbuild.conf, I get the very same error.[/quote]

I believe this method is correct. However, based on the error message you shared, it looks like BOOT_SWAP_USING_MOVE is not recognized in the sysbuild.conf.

Please check the build configuration files to see if the configuration values have been propagated correctly. Make sure to check both the mcuboot build config and the application build config files.

Kind Regards,

Abhijith