How to update the MCUboot in nRF Connect on nRF52840DK

RE: How to update the MCUboot in nRF Connect on nRF52840DK

Sigurd — Thu, 17 Dec 2020 21:59:05 GMT

Hi Ram,

Try setting it to the EC key instead.

For me it's:

CONFIG_SB_SIGNING_KEY_FILE="C:/ncs/bootloader/mcuboot/root-ec-p256.pem"

I don't think you need to set CONFIG_SB_PUBLIC_KEY_FILES (I think it will be generated based on the private/signing key for you, but if you are setting it manually, it need to generated based on the private key).

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Thu, 17 Dec 2020 06:09:12 GMT

Hi Sigurdon,

I made the second time build without pristine and without deleting the build directory and used _b0_s0_image_signed.hex to program the bootloader. It worked and printed the new message when put in the DFU mode. Thank you very much for pointing me to the problem!

However, I am getting build error with CONFIG_SB_SIGNIG_KEY_FILE.

This is what I put in the blinky's prj.conf file.

CONFIG_SB_SIGNING_KEY_FILE="c:/Users/singhr48/ncs/v1.4.0/bootloader/mcuboot/root-rsa-2048.pem"
CONFIG_SB_PUBLIC_KEY_FILES="c:/Users/singhr48/ncs/v1.4.0/bootloader/mcuboot/enc-rsa2048-pub.pem"

The two .pem files are provided as part of the ncs.

And the build error is

[150/162] cmd.exe /C "cd /D C:\Users\singhr48\ncs\v1.4.0\zephyr\samples\basic\blinky\build\zephyr && C:\Users\singhr48\ncs\v1.4.0\toolchain\opt\bin\python.exe C:/Users/singhr48/ncs/v1.4.0/nrf/scripts/bootloader/keygen.py --public --in c:/Users/singhr48/ncs/v1.4.0/bootloader/mcuboot/root-rsa-2048.pem --out C:/Users/singhr48/ncs/v1.4.0/zephyr/samples/basic/blinky/build/zephyr/nrf/subsys/bootloader/generated/public.pem"
Traceback (most recent call last):
File "C:/Users/singhr48/ncs/v1.4.0/nrf/scripts/bootloader/keygen.py", line 33, in <module>
sk = SigningKey.from_pem(args.infile.read()) if args.infile else SigningKey.generate(curve=NIST256p)
File "C:\Users\singhr48\ncs\v1.4.0\toolchain\opt\bin\lib\site-packages\ecdsa\keys.py", line 828, in from_pem
privkey_pem = string[string.index(b("-----BEGIN EC PRIVATE KEY-----")):]
ValueError: subsection not found
FAILED: zephyr/nrf/subsys/bootloader/generated/public.pem
cmd.exe /C "cd /D C:\Users\singhr48\ncs\v1.4.0\zephyr\samples\basic\blinky\build\zephyr && C:\Users\singhr48\ncs\v1.4.0\toolchain\opt\bin\python.exe C:/Users/singhr48/ncs/v1.4.0/nrf/scripts/bootloader/keygen.py --public --in c:/Users/singhr48/ncs/v1.4.0/bootloader/mcuboot/root-rsa-2048.pem --out C:/Users/singhr48/ncs/v1.4.0/zephyr/samples/basic/blinky/build/zephyr/nrf/subsys/bootloader/generated/public.pem"
ninja: build stopped: subcommand failed.

Could you please guide what I am missing while defining the key files?

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Wed, 16 Dec 2020 16:27:28 GMT

Thanks your reply!

Yes, I always delete the build directory and do pristine build.

I am not using my own key file to sign the build, and this may be the reason as explained by you. I will use a predefined key file with CONFIG_SB_SIGNIG_KEY_FILE and repeat the test.

Would the sample key file provided with ncs work?

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

Sigurd — Wed, 16 Dec 2020 15:58:45 GMT

Hi Ram,

[quote userid="96121" url="~/f/nordic-q-a/68787/how-to-update-the-mcuboot-in-nrf-connect-on-nrf52840dk/284855#284855"]What may be the reason the signature of the bootloaders (_b0_s0_ and _b0_s1_) is changed in a built with a small change?[/quote]

Did you do a pristine build (e.g. west build with -p argument), or deleted the old build folder, when you build the project with the change(a modified printf statement)? If you are not setting your own CONFIG_SB_SIGNING_KEY_FILE, then the generated debug keys will be used. New debug keys are generated if you delete the build folder, or do a pristine build.

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Mon, 14 Dec 2020 15:56:06 GMT

Hi,

I think my earlier observation was not correct, and update always happens at offset 0x200.

But, I am not able to update the bootloader using the nRF programmer. This is what I have found while programming the device with the nRF programmer.

1. Build the project and program the device (b0, mcuboot in so and s1, and application). I used merged.hex

2. Build the project without any change, and program the device with _b0_s0 (OR _b0_s1_, OR app_update.hex), no problem.

3. Build the project with a small change (modify a printf statement), and program the device with _b0_s0. The device fails to boot with _b0_s0 but boots with _b0_s1. I have found that the bootloader signature is changed.

4. Program the device with the newly built app. There is no issue.

What may be the reason the signature of the bootloaders (_b0_s0_ and _b0_s1_) is changed in a built with a small change? I really appreciate your insights.

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Thu, 03 Dec 2020 16:59:37 GMT

Hi,

Please find attached pm.config and partition. yml

Is the address a reason for the mcuboot update not working with the programmer?

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Thu, 03 Dec 2020 16:57:18 GMT

EMPTY_0:
address: 0x18200
placement:
before:
- s1_pad
region: flash_primary
size: 0xe00
EMPTY_1:
address: 0x19200
placement:
before:
- s1_image
region: flash_primary
size: 0xe00
app:
address: 0x2a200
region: flash_primary
size: 0x6ae00
b0:
address: 0x0
orig_span: &id001
- b0_image
- provision
region: flash_primary
size: 0x8000
span: *id001
b0_image:
address: 0x0
placement:
after:
- start
region: flash_primary
size: 0x7000
mcuboot:
address: 0x8200
placement:
before:
- mcuboot_primary
region: flash_primary
sharers: 0x1
size: 0x10000
mcuboot_pad:
address: 0x2a000
placement:
align:
start: 0x1000
before:
- mcuboot_primary_app
region: flash_primary
sharers: 0x2
size: 0x200
mcuboot_primary:
address: 0x2a000
orig_span: &id002
- app
- mcuboot_pad
region: flash_primary
sharers: 0x1
size: 0x6b000
span: *id002
mcuboot_primary_app:
address: 0x2a200
orig_span: &id003
- app
region: flash_primary
size: 0x6ae00
span: *id003
mcuboot_secondary:
address: 0x95000
placement:
after:
- mcuboot_primary
align:
start: 0x1000
region: flash_primary
share_size:
- mcuboot_primary
size: 0x6b000
provision:
address: 0x7000
placement:
after:
- b0_image
align:
start: 0x1000
region: flash_primary
size: 0x1000
s0:
address: 0x8000
orig_span: &id004
- mcuboot
- s0_pad
region: flash_primary
size: 0x10200
span: *id004
s0_image:
address: 0x8200
orig_span: &id005
- mcuboot
region: flash_primary
size: 0x10000
span: *id005
s0_pad:
address: 0x8000
placement:
after:
- b0
align:
start: 0x1000
region: flash_primary
share_size:
- mcuboot_pad
size: 0x200
s1:
address: 0x19000
orig_span: &id006
- s1_pad
- s1_image
region: flash_primary
size: 0x10200
span: *id006
s1_image:
address: 0x1a000
placement:
after:
- s1_pad
- s0
align:
end: 0x1000
region: flash_primary
share_size:
- mcuboot
size: 0x10000
s1_pad:
address: 0x19000
placement:
after:
- s0
align:
start: 0x1000
region: flash_primary
share_size:
- mcuboot_pad
size: 0x200
spm_app:
address: 0x2a200
orig_span: &id007
- app
region: flash_primary
size: 0x6ae00
span: *id007
sram_primary:
address: 0x20000000
region: sram_primary
size: 0x40000

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Thu, 03 Dec 2020 16:57:00 GMT

devzone.nordicsemi.com/.../pm_2E00_config

RE: How to update the MCUboot in nRF Connect on nRF52840DK

Sigurd — Thu, 03 Dec 2020 13:11:54 GMT

Hi,

It's the Partition Manager that sets the start address in a multi-image build. See this link for more information: http://developer.nordicsemi.com/nRF_Connect_SDK/doc/1.4.0/nrf/scripts/partition_manager/partition_manager.html

In your build folder, how does partitions_nrf52840dk_nrf52840.yml and pm_nrf52840dk_nrf52840.config look like ?

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Wed, 02 Dec 2020 19:45:52 GMT

Thank you very much!

I hope there will be solution soon.

One more question. If I program signed_by_mcuboot_and_b0_s0_image_signed.hex (or _b0_s1_) using the nRF programmer, it should work because the hex file has the address. But, it is not working.

I have noted that the signed_by_mcuboot_and_b0_s0_image_signed.hex (or _b0_s0_) is loaded at 0x8000 (0x19000), not at 0x8200 (0x19200) as expected. The pp_signed.hex is loaded at 0x2A200.

Is the address a reason for the mcuboot update not working with the programmer? Overall image size has not changed (verified by reading the image in the programmer)

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

Sigurd — Wed, 02 Dec 2020 18:50:17 GMT

Hi,

Unfortunately, DFU of the MCUBoot is currently not supported by mcumgr. You are correct, it doesn't accept an address or slot parameter, and is initialized with the application address. There are some related activity ongoing around this (- mcumgr/SMP support for recognizing which image need to be upload (to be defined)) in this PR: github.com/.../27673

[quote userid="96121" url="~/f/nordic-q-a/68787/how-to-update-the-mcuboot-in-nrf-connect-on-nrf52840dk"]Are there different ways to enter the DFU mode to update the application and to update the mcuboot?[/quote]

Unfortunately, I don't think there are other options/tools available. Sorry for the inconvenience.

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Tue, 01 Dec 2020 15:23:11 GMT

Hi,

Any update on the mcuboot update and method of entering the DFU mode for it is well appreciated.

Thanks,

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

RAS_ID — Mon, 30 Nov 2020 07:59:55 GMT

Hi,

Thanks for your reply!

I use the following command to program the flash.

$ mcumgr --conntype serial --connstring="COM14" image upload build/zephyr/signed_by_mcuboot_and_b0_s1_image_update.bin -mcumgr command used to program the flash.

63.82 KiB / 63.82 KiB 100.00% 2.92 KiB/s 21s01s -the numbers changes during programming
Done -once programming is done.

and, application prj.conf is given below with default prj.conf for the mcuboot

CONFIG_GPIO=y
CONFIG_SECURE_BOOT=y
CONFIG_BOOTLOADER_MCUBOOT=y
CONFIG_BUILD_S1_VARIANT=y

Memory partition looks like as given below

immutable bootloader starts at 0

S0 starts at 8000

S1 starts at 19000

App starts at 2A000

When I compare the files read after programming, I see the following

After programming with app_update.bin, no difference except at the end of the file.

After programming with signed_by_mcuboot_and_b0_s1_image_update.bin, the differences start at 2A000. I also recorded that the read file size is 100KB larger than the original file.

Please let me know if I could provide any other details.

Thanks

Ram

RE: How to update the MCUboot in nRF Connect on nRF52840DK

Sigurd — Fri, 27 Nov 2020 14:51:17 GMT

Hi,

I find it strange that the mcuboot update affected the application area in that way.

What command did you use with mcumgr, and how does your prj.conf file look like ?

Did you try with signed_by_mcuboot_and_b0_s0_image_update.bin as well?

What does the mcumgr image list command return?