dist/testbed-support: add reset after flash for IoT-LAB nodes

[baptleduc]

makefile.iotlab.single.inc.mk does not define flash-recipe, so it falls back to the default defined in Makefile.include:832:

define default-flash-recipe
  $(call check_cmd,$(FLASHER),Flash program)
  $(PROGRAMMER_FLASH)
endef
flash-recipe ?= $(default-flash-recipe)

This default doesn't reset the board after flashing. As a result, the node does not restart automatically and produces no output until a manual reset is triggered.

This was observed on b-l072z-lrwan1 nodes at the IoT-LAB Saclay site: examples/basic/hello-world produced no serial output after make flash term until the node was reset manually.

Fix: define an iotlab-flash-recipe that appends a reset step after flashing.

Testing procedure

Tested on b-l072z-lrwan1 at IoT-LAB Saclay. After flashing, the node now resets automatically and serial output appears immediately without manual intervention.

Issues/PRs references

Originally posted in #22271 (comment)

Declaration of AI Tools / LLMs usage

• Claude Code for code review and guidance, with user authorship of the implementation

[riot-ci]

Murdock results

✔️ PASSED

def4576 dist/testbed-support: add flash-recipe to iotlab

Success	Failures	Total	Runtime
1	0	1	01m:21s

Artifacts

• Documentation preview

[AnnsAnns]

🤔 I wonder whether this also fixes some release test weirdness with iotlab

[crasbe]

🤔 I wonder whether this also fixes some release test weirdness with iotlab

There is only one way to find out... Do another release! 😈

[baptleduc]

🤔 I wonder whether this also fixes some release test weirdness with iotlab

There is only one way to find out... Do another release! 😈

Could I help you in any way ?

[crasbe]

🤔 I wonder whether this also fixes some release test weirdness with iotlab

There is only one way to find out... Do another release! 😈

Could I help you in any way ?

That was just a joke. The release tests can be run without doing a release too :D

[baptleduc]

🤔 I wonder whether this also fixes some release test weirdness with iotlab

There is only one way to find out... Do another release! 😈

Could I help you in any way ?

That was just a joke. The release tests can be run without doing a release too :D

Oops 😅😂

But do you want me to open a PR in te iot lab repo ?

[baptleduc]

Hello,
Is it possible to have an update please ?
Thanks in advance

[crasbe]

Unfortunately I have no experience with the IoT-LAB, but I pinged the other maintainers to take a look.

[aabadie]

I tried to flash a RIOT firmware on a b-l072z-lrwan1 board of the IoT-LAB Saclay site. Here is the output:

$ make BOARD=b-l072z-lrwan1 -C examples/basic/default/ flash term IOTLAB_NODE=auto
make: Entering directory '/work/riot/RIOT/examples/basic/default'
Building application "default" for "b-l072z-lrwan1" with CPU "stm32".

[...]

   text	   data	    bss	    dec	    hex	filename
  43164	    124	   6320	  49608	   c1c8	/work/riot/RIOT/examples/basic/default/bin/b-l072z-lrwan1/default.elf
iotlab-node --jmespath='keys(@)[0]' --format='lambda ret: exit(int(ret))'  --list saclay,st-lrwan1,10 --flash /work/riot/RIOT/examples/basic/default/bin/b-l072z-lrwan1/default.bin
ssh -t abadie@saclay.iot-lab.info 'socat - tcp:st-lrwan1-10.saclay.iot-lab.info:20000' 
main(): This is RIOT! (Version: 2025.10-devel-463-gce692)
Welcome to RIOT!
> help
help
Command              Description
---------------------------------------
ifconfig             Configure network interfaces
txtsnd               Sends a custom string as is over the link layer
pm                   interact with layered PM subsystem
ps                   Prints information about running threads.
rtc                  control RTC peripheral interface
saul                 interact with sensors and actuators using SAUL
version              Prints current RIOT_VERSION
reboot               Reboot the node
> ^CConnection to saclay.iot-lab.info closed.

As you can see the firmware starts automatically. So I insist that in your case (you are testing with the "hello-world" application), what you are experiencing is that the firmware has already finished executing when the socat terminal client connect to the serial output.
Just add a "ztimer_sleep(ZTIMER_SEC, 5)" at the beginning of main and add the ztimer dependency in the makefile, and you'll the hello world coming after the 5s delay.

[aabadie]

Here is the openocd code called when flashing a firmware on those devices in iot-lab: https://github.com/iot-lab/iot-lab-gateway/blob/71ab066ea3229cdcd4be22a5b8e43910888293ef/gateway_code/utils/openocd.py#L72-L79
As you can see the there's a call to "reset run" after the firmware is written on the flash.

[aabadie]

BTW, the output above was obtained from the master branch of RIOT (to avoid confusion)

[baptleduc]

Thank you very much for your explanation.
I think we can, therefore, close this PR.

[AnnsAnns]

Thank you for testing this @aabadie and also thank you for initially trying to solve this issue @baptleduc 👍