nrf52840 (pca10059) crashed at the very simplest program

[hmaerki]

CircuitPython version and board name

boot_out.txt:
Adafruit CircuitPython 10.0.3 on 2025-10-17; PCA10059 nRF52840 Dongle with nRF52840
Board ID:pca10059
UID:AC2D4B6D80A68DEA
boot.py output:

boot.py:
import storage
import supervisor
storage.disable_usb_drive()
supervisor.runtime.autoreload = True

Code/REPL

x  # Yes - this program makes pca10059 crash!

Behavior

No response, board hangs and has to be powercycled

Description

Expected behaviour

>>> x
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'x' is not defined

This expected behavior is observed on:

• pca10059 9.2.9
• Seeed XIAO nRF52840 Sense with nRF52840, 9.2.9, 10.0.1, 10.0.2, 10.0.3

However the crash is observed on

• pca10059 10.0.1, 10.0.2, 10.0.3

Additional information

No response

[dhalbert]

If you remove one or more of the lines from boot.py, does it make any difference?

[hmaerki]

I left boot.py how it was - but here is a sequence with the expected output and a crash:

First start with a well defined filesystem

>>> import storage; storage.erase_filesystem()
-> need powercycle

Copy boot.py via usb mount

$ cat boot.py
import storage
import supervisor
storage.disable_usb_drive()
supervisor.runtime.autoreload = False
$ cp boot.py /media/maerki/CIRCUITPY/

Expected behaviour

Powercycle

>>> x
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'x' isn't defined

Now crash

Soft-reboot with <ctrl+d> will make it crash

>>> x
CPU hangs

When I do the same with Adafruit CircuitPython 10.0.3 on 2025-10-17; Seeed XIAO nRF52840 Sense with nRF52840, it will NOT crash!

Changing the contents of boot.py

The crash happens even if boot.py is empty!

Observation

Not 100% shure: Always after <ctrl+d> and then >>> x, the pca10059 crashes and needs to be powercycled.

[dhalbert]

I have a couple of PCA10059's on one. I tried 10.0.3 on one, and could not reproduce your error.

When did you get the PCA10059, and did you flash the UF2 bootloader onto the board? Or did you try loading CircuitPython a different way?

If you flashed the UF2 bootloader a long time ago, I am suspicious that it is out of date. For recent versions of CircuitPython, you really need a newer bootloader to accomodate the larger amount of firmware space that CircuitPython now uses. Normally the UF2 would not work at all, but it might just work on this board.

Look at the first version number in the INFO_UF2.TXT file shown in NRF52BOOT. My guess is that it is fairly old, depending on when you flashed the bootloader. See this page https://learn.adafruit.com/introducing-the-adafruit-nrf52840-feather/update-bootloader. Note that there are three ways to update the bootloader, and that the UF2 method does not work if the bootloader is older than a certain version. It's all explained on that page and the following pages.

If you update with the adafruit-nrfutil method, use this zip file to update: https://github.com/adafruit/Adafruit_nRF52_Bootloader/releases/download/0.9.2/pca10059_bootloader-0.9.2_s140_6.1.1.zip. You do not need to unzip the file.

[hmaerki]

Hi Dan

Thank you very much for your time and precise responses so far.
I have two dongles, both very new. I always flashed the bootloader using picoprobe and never used something older than 0.9.2. Something is leading us astray...
The second dongle was brand new and below you see what I did do it. Conclusion: Exactly the same error. But rarely 10.0.3 gives a correct stack trace. It has something to do with the boot sequence - there must be some state around in the registers or the filesystem. 9.2.9 nverver crashes.

Here in sequence what I dit with the second dongle ('x23).

Flaky test on second nrf52840 USB dongle

Tag	Address	hw version	Command
first	E2:D8:94:F7:96:91	2.1.1, 2024.2	Here I observed the flakyness
second	DC:FD:03:56:44:24	2.1.1, 2023.31	Used to verify behaviour

first

cat NRF52BOOT/INFO_UF2.TXT
Model: Nordic nRF52840 Dongle
Board-ID: nRF52840-pca10059-v1
Date: Jul 19 2024
SoftDevice: S140 6.1.1

cat CIRCUITPY/boot_out.txt
Adafruit CircuitPython 10.0.3 on 2025-10-17; PCA10059 nRF52840 Dongle with nRF52840
Board ID:pca10059
UID:AC2D4B6D80A68DEA
boot.py output:

second

cat NRF52BOOT/INFO_UF2.TXT
Model: Nordic nRF52840 Dongle
Board-ID: nRF52840-pca10059-v1
Date: Jul 19 2024
SoftDevice: S140 6.1.1

cat CIRCUITPY/boot_out.txt 
Adafruit CircuitPython 10.0.3 on 2025-10-17; PCA10059 nRF52840 Dongle with nRF52840
Board ID:pca10059
UID:9CA607F4328D2568

Flashed bootloader using Picoprobe

wget https://github.com/adafruit/Adafruit_nRF52_Bootloader/releases/download/0.9.2/pca10059_bootloader-0.9.2_s140_6.1.1.hex
export IMAGE=pca10059_bootloader-0.9.2_s140_6.1.1.hex
openocd -f interface/cmsis-dap.cfg -f target/nordic/nrf52.cfg -c "gdb_flash_program enable" -c "init; nrf52_recover; program "$IMAGE" verify reset exit"

Loaded circitpython 10.0.3 using

(cd /media/maerki/NRF52BOOT/; wget https://adafruit-circuit-python.s3.amazonaws.com/bin/pca10059/en_US/adafruit-circuitpython-pca10059-en_US-10.0.3.uf2)

Just let it reboot and enter:

mpremote exec 'x'
--> Crash!

Update bootloader using UF2

(cd /media/maerki/NRF52BOOT/; wget https://github.com/adafruit/Adafruit_nRF52_Bootloader/releases/download/0.9.2/update-pca10059_bootloader-0.9.2_nosd.uf2)

mpremote exec 'x'
--> Crash!

Format filesystem

>>> import storage
>>> storage.erase_filesystem()

mpremote exec 'x'
--> Crash!

Loaded circitpython 9.2.9 using

(cd /media/maerki/NRF52BOOT/; wget https://adafruit-circuit-python.s3.amazonaws.com/bin/pca10059/en_US/adafruit-circuitpython-pca10059-en_US-9.2.9.uf2)

Just let it reboot and enter:

mpremote exec 'x'
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'x' is not defined
--> success

Loaded circitpython 10.0.3 using

(cd /media/maerki/NRF52BOOT/; wget https://adafruit-circuit-python.s3.amazonaws.com/bin/pca10059/en_US/adafruit-circuitpython-pca10059-en_US-10.0.3.uf2)

mpremote exec 'x'
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'x' is not defined
--> success (first success on 10.0.3...)

mpremote exec 'x'
--> Crash!

Additional Comments

I observed the crash in Thonny using the REPL and mpremote exec 'x'.

[dhalbert]

Could you try it again without using mpremote? Just connect to the REPL via a terminal program (we like tio) and type x at the >>> prompt.

Both Thonny and mpremote do tricky things to execute stuff in the REPL. They go into paste or raw mode, etc. mpremote is also not tested to work with CircuitPython, only MicroPyton.

[hmaerki]

Hi Dan

$ tio /dev/ttyACM0
Adafruit CircuitPython 10.0.3 on 2025-10-17; PCA10059 nRF52840 Dongle with nRF52840
>>> x
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'x' isn't defined

The crash has gone! And my problem is solved so far. Thank you very much!

Do you want me to investigate why the nrf52840 crashes (sniff at the UART traffic)?

[dhalbert]

I think this may be another manifestation of #8038 or #10665. Earlier investigation there showed that an unhandled exception while in paste or maybe raw mode can cause a crash. But a log of what mpremote is sending and what happens would be great. I wonder if mpremote has some debug logging option for that.

[hmaerki]

But a log of what mpremote is sending and what happens would be great.
I trace it and send you a log tomorrow. If I do not find a trace utility for mpremote I will tweak pyserial.

[hmaerki]

Hi Dan

Here are the logs. Please tell me if I can support you more on this topic.

mpremote with debug output of raw ascii

I patched mpremote so it allows us to see all characters on the serial line.

mpremote tries to bring the device in a specific state (raw mode, soft reset, raw paste mode). I added debug output whenever such a mode is entered.

This mpremote may be helpful to debug #8038 and #10665 too.

Testresults

I use uv on ubuntu (not tested on windows/max).

uv run --with=git+https://github.com/hmaerki/fork_micropython@hmaerki_circuitpython_issue_10717#subdirectory=tools/mpremote mpremote exec x >& filename.txt

• pca10059_9.2.9.txt circitpython pca10059 9.2.9 which works ok:

• pca10059_10.0.3.txt circitpython pca10059 10.0.3 which results in a crash

Observation

In both logs, you see MPREMOTE: raw_paste_write(): Indicate end of data.. Just after this, the device crashes.

[bablokb]

Raw paste mode is very fragile. I have a project that uses basically the same low-level transport mechanisms as mpremote and depending on the device you really need to transfer data using very small chunks.